Since its emergence as a unified state in 1747, Afghanistan has known almost no sustained period of internal peace. In under two centuries, it has fought five major wars against external powers: three against Britain (1839-1842, 1878-1880, 1919), one against the Soviet Union (1979-1989), and one against a US-led coalition (2001-2021). This is a frequency of great-power confrontation matched by few other states in the modern international system. Each of these conflicts has been the subject of extensive historical treatment ([1]; [8]). What has received far less systematic attention is the relationship between them: whether these five wars are best understood as discrete historical episodes, each explicable on its own terms, or as stages in a single, causally connected sequence. This article argues for the latter, and asks a correspondingly simple question: why have Afghanistan’s conflicts persisted for so long, and how has each stage shaped the conditions for the next?

The question is not merely historical. The 2021 withdrawal of US and NATO forces and the Taliban’s return to power reopened it in acute form. Four years on, the Islamic Emirate of Afghanistan remains formally unrecognized by every state but Russia ([12]), while cross-border fighting between Afghanistan and Pakistan escalated from sporadic skirmishing in late 2025 to open conflict by February 2026, with Pakistani strikes on Kabul and Kandahar and well over 100,000 people displaced ([11]). Whether this represents a new equilibrium in Afghan politics or the leading edge of a sixth stage in a much older pattern is a question this article returns to in its concluding sections. But it cannot be answered without first establishing what that pattern is.

The existing literature tends to approach Afghanistan’s conflicts through one of two lenses, rarely both. A first body of work, largely historical and area-studies in orientation, treats each war as a product of its specific moment: the nineteenth-century Anglo-Russian rivalry explains the first three wars ([28]; [20]), Cold War bipolarity explains the Soviet invasion ([3]), and post-9/11 counterterrorism policy explains the American intervention ([17]). A second body of work, more common in security studies, emphasizes Afghanistan’s internal political structure, meaning its decentralized tribal and ethnic configuration, weak state institutions, and contested political legitimacy, as the primary obstacle to durable peace, largely independent of external involvement ([1]). Both literatures are empirically rich, but their analytical separation obscures a pattern that becomes visible only when the two are read together: that external intervention and internal fragmentation are not competing explanations for Afghanistan’s conflicts but interacting ones, each intensifying the conditions that produce the other.

This article develops that argument through a neoclassical realist framework. Structural, or neorealist, accounts explain why great powers have repeatedly found Afghanistan strategically important despite its limited intrinsic value to them: it functioned as a buffer state between rival empires and, later, rival blocs ([27]). What neorealism at the systemic level cannot explain is why identical external pressure produced markedly different outcomes across Afghanistan’s five conflicts, or why internal Afghan actors so often shaped those outcomes independently of great-power intent. Neoclassical realism addresses this gap by treating domestic-level variables, namely state capacity, elite cohesion, and regime legitimacy, as intervening variables that filter systemic pressure before it produces a given foreign policy or conflict outcome ([21]; [19]). Applied to Afghanistan across two centuries, this framework suggests that each war left both external dependency and internal fragmentation unresolved, and that this unresolved pair of conditions became, in turn, the precondition for the next external intervention.

The article makes three contributions. First, it offers a five-stage framework linking the Anglo-Afghan wars, the Soviet invasion, and the American occupation into a single causal sequence, rather than treating them as isolated historical episodes. Second, it uses that framework to analyze systematically how external and internal factors interact, rather than merely coexist, in sustaining Afghan conflict, which is the article’s central analytical claim. Third, it extends the analysis to the post-2021 period, including the still-unfolding Afghanistan-Pakistan conflict, to ask whether the same structural pattern is reasserting itself under Taliban rule. In doing so, the article aims to move the study of Afghan conflict beyond a choice between externalist and internalist explanations, toward an account of how the two have sustained one another since 1839.

The remainder of the article proceeds as follows. Section 2 develops the theoretical framework, situating neoclassical realism relative to structural neorealism and to [26] levels-of-analysis approach. Section 3 synthesizes the five conflicts into a comparative historical overview, identifying the recurring patterns of buffer-state dependency, proxy militarization, and unresolved internal settlement that connect them. Section 4 undertakes the article’s core analysis, examining external and internal factors in turn before addressing their interplay directly. Section 5 extends this analysis to the conditions for peace since 2021, including the current Afghanistan-Pakistan conflict. Section 6 concludes with the article’s broader theoretical implications and a small set of policy recommendations that follow directly from the framework developed here.

Structural neorealism offers the most parsimonious starting point for explaining why great powers have repeatedly gone to war over Afghanistan. In [27] account, the anarchic structure of the international system compels states to attend to relative power and security regardless of any particular state’s intrinsic value; a state can become strategically significant simply by occupying contested space between rivals. Afghanistan’s geography made it exactly this kind of space. It functioned first as the buffer separating British India from Tsarist Russia during the nineteenth-century “Great Game”, and later as a Cold War frontier between the Soviet Union and a US-aligned Pakistan. On this reading, the recurrence of external intervention in Afghanistan is not a puzzle about Afghanistan at all; it is an ordinary consequence of great-power competition playing out over strategically located terrain.

This structural account, however, has clear limits. Neorealism explains why great powers found Afghanistan worth contesting, but it has much less to say about why the same underlying pressure produced such different trajectories across cases: swift British defeat in 1842, a costly but partial British victory in 1880, full Afghan independence after 1919, a decade-long Soviet counterinsurgency that ultimately failed, and a two-decade American state-building project that collapsed within weeks of withdrawal. [21] identifies this as the central shortcoming of purely systemic theory: because structural variables are held constant across cases in ways that domestic variables are not, a theory confined to the international level cannot account for variation in how states translate systemic pressure into foreign policy outcomes.

Neoclassical realism supplies the missing piece. As formulated by [21] and elaborated by [19], the theory treats systemic pressure as necessary but not sufficient to explain outcomes; that pressure must pass through domestic-level intervening variables, including state capacity, elite cohesion, and the coherence of national institutions, before it produces a given result. [19] develop “the state” itself as an analytical category in this process: a strong, cohesive state can absorb or redirect external pressure, while a fragmented one is instead reshaped by it, often in ways that generate further instability. This is the theoretical move this article makes central to its analysis of Afghanistan: identical external pressure, whether invasion, proxy support, or occupation, has produced divergent outcomes across the five conflicts in large part because Afghanistan’s internal cohesion at each historical moment differed substantially.

[26] earlier work in Man, the State, and War provides the organizing scaffold for the analysis that follows. Waltz’s three “images”, the individual, the state, and the international system, offer a way of sorting causal claims about war that this article adapts into a simpler external/internal distinction: external factors corresponding to Waltz’s systemic image, and internal factors corresponding to his second image of the state’s internal composition. Section 4 of this article is structured explicitly around this distinction, first examining external factors, then internal factors, before turning, in the article’s most original contribution, to their interplay.

A rentier-state account offers the most direct rival explanation to the one developed here, and is worth engaging on its own terms rather than merely noting in passing. [22] argues that Afghan political elites since the nineteenth century have built and maintained power chiefly through external subsidies: British stipends to nineteenth-century rulers, Soviet and American aid during the Cold War, and internationally financed budgets after 2001, rather than through domestic taxation and accountability to Afghan society. On this account, Afghanistan’s conflicts are best explained not by the interaction of external leverage and internal fragmentation developed here, but more simply by the collapse of external rents: when a patron’s subsidy is reduced or withdrawn, the rentier state built on it collapses, and conflict follows almost mechanically. This is a serious rival hypothesis, and it accounts for a good deal of the pattern documented in Section 3. Most starkly, it accounts for the collapse of the Najibullah government within months of the final suspension of Soviet subsidies in 1991-1992 ([23]), and, arguably, the collapse of the Republic within weeks of the announced 2021 US withdrawal. What a purely rentier account cannot explain, however, is why external rents produced state collapse in some periods but not others of comparable subsidy dependence, nor why, once rents were withdrawn, some factions collapsed almost immediately (the PDPA leadership after 1989, the Republic’s security forces after 2021) while others converted the same external funding into durable local power bases that outlasted their patrons entirely (the Mujahideen commanders who became warlords after 1992; [7]). This variation in how actors respond to identical rent shocks is precisely what the intervening-variable logic of neoclassical realism is built to explain, and a purely rentier account, which treats the Afghan state as a single undifferentiated recipient of rents, is not equipped to explain it.

As a matter of research design, this article’s scope condition should be stated explicitly rather than left implicit. The five stages developed here are defined by direct military intervention by an external great power: Britain three times, the Soviet Union once, and the United States once. This is a deliberate restriction rather than an incidental one. It excludes internally generated conflicts without comparable direct great-power military involvement, including the 1929 civil war that briefly toppled King Amanullah’s successor and the 1978 Saur Revolution. This restriction keeps the comparison analytically tractable: each of the five cases involves a comparable class of event, namely foreign invasion or occupation, which is what makes systematic comparison across nearly two centuries possible in the first place. It does, however, create one significant analytical cost, which Section 3 addresses directly: the 1989-2001 period falls outside the strict scope condition, since no external power directly invaded Afghanistan during it, yet it is causally indispensable to the argument, because it is the period in which the unresolved legacies of the Soviet invasion and Afghanistan’s internal fragmentation together produced the Taliban regime that the 2001 invasion then targeted. Section 3 therefore treats this period as a bridging interlude: analytically distinct from the five external-intervention stages, but essential to connecting stage four to stage five.

These competing expectations can be stated case by case rather than only in general terms. For the Soviet period, a rentier account predicts that the Najibullah government’s fate should track the Soviet subsidy alone, collapsing when and only when the subsidy was cut; neoclassical realism predicts that its fate should also depend on whether the government could convert that subsidy into a cohesive ruling coalition, which by the late 1980s it manifestly could not, given the PDPA’s persistent Khalq-Parcham factionalism ([23]; [6]). For the post-2001 period, a rentier account predicts that the Republic’s collapse should track the withdrawal of US financing; neoclassical realism predicts, in addition, that the manner of its collapse, security forces dissolving without sustained resistance rather than negotiating a transition, should reflect the Republic’s failure to build a cohesive officer corps and political elite independent of the aid relationship itself ([17]). Evidence for the interaction mechanism beyond the subsidy relationship itself is visible in exactly this variation: Mujahideen commanders who received comparable external funding to the PDPA nonetheless converted it into durable, personally loyal power bases that outlasted their patrons entirely ([7]), while the PDPA leadership and, later, the Republic’s leadership did not. The difference lies not in the size or timing of external support but in the domestic political structures receiving it, which is precisely the intervening variable a rentier account leaves unexamined.

To keep this comparison analytically consistent rather than impressionistic, each of the five stages developed below is assessed along the same five dimensions, coded from the secondary sources cited in Section 3: external leverage (the type and depth of the external power’s involvement, from direct military occupation to indirect subsidy and arms supply); state capacity (the extent to which the incumbent or installed Afghan government could raise domestic revenue, administer territory beyond Kabul, and maintain an army loyal to the central government rather than to a faction or patron); elite cohesion (whether the ruling coalition held together as a single bargaining unit or split into rival factions with independent bases of power); legitimacy (whether the government’s claim to rule was broadly accepted domestically and internationally, or actively contested on ideological, ethnic, or dynastic grounds); and conflict outcome (whether the stage ended in the incumbent government’s survival, its collapse and replacement, or a negotiated settlement). Applying these five dimensions consistently across cases, rather than assessing each stage on its own terms, is what allows Section 3’s comparative table and Section 4’s analysis to identify patterns rather than simply narrate five separate histories.

This framework has further limits, which are worth stating plainly. Beyond the rentier-state account addressed above, other frameworks could tell complementary stories about the same events: a constructivist account might foreground the role of Islamic identity and legitimacy in ways this article touches on only briefly, while a historical-institutionalist account might place greater weight on path-dependent administrative structures inherited from earlier periods ([1]). The analysis that follows is methodologically qualitative and interpretive rather than a formal test of competing hypotheses, and draws primarily on secondary historical and policy sources rather than original archival research. This article does not claim that neoclassical realism is the only lens through which Afghanistan’s conflicts can be understood, only that, against its most direct rival, it better explains the variation in outcomes documented across the cases that follow.

Rather than narrate each of Afghanistan’s five wars in full, this section synthesizes them comparatively, using Table 1 to establish the empirical pattern that Section 4 then explains.

Table 1. The five stages of Afghan conflict and their unresolved legacies, 1839-2021.

Three patterns cut across these five cases. The first is buffer-state dependency: each Anglo-Afghan war reinforced Afghanistan’s position between rival empires, so that external dependency became a structural feature of Afghan statehood rather than a one-time condition imposed from outside ([28]; [1]). The second is proxy militarization: from the Anglo-Russian rivalry of the nineteenth century through Cold War competition between the Soviet Union and the US-Pakistan axis, each stage armed and empowered internal Afghan factions as instruments of external rivalry, leaving those factions in place, and often in conflict with one another, once the sponsoring power withdrew ([5]; [18]). The third is unresolved internal settlement: none of the five wars concluded with a durable internal power-sharing arrangement or a state monopoly on organized violence, which is precisely the condition that neoclassical realism identifies as making a state vulnerable to renewed external shock ([19]).

These patterns can also be traced stage by stage as explicit causal links, each with its own intervening mechanism and supporting source, rather than inferred from chronological succession alone. The passage from the first to the second Anglo-Afghan war ran through Britain’s persistent “forward policy” toward Afghan foreign relations: the same anxiety about Russian influence that had driven the 1839 invasion resurfaced when Amir Sher Ali received a Russian diplomatic mission in 1878, prompting the British ultimatum that began the second war ([15]; [28]). The passage from the second to the third war ran through the Treaty of Gandamak itself: the treaty’s cession of Afghan foreign-policy sovereignty to Britain was the specific grievance Amanullah Khan’s 1919 declaration of independence was designed to reverse ([8]). The passage from the third war to the Soviet invasion is less direct but still traceable through a documented chain of domestic political events: the personalist, institutionally thin monarchy that Afghan independence left in place proved unable to manage its own succession, leading to Daoud’s 1973 coup, the 1978 Saur Revolution, and a fragile PDPA government whose instability the Soviet leadership judged serious enough to warrant direct intervention in 1979 ([1]). The passage from the Soviet invasion through the bridging interlude to the American invasion is documented in the greatest detail of the four transitions, and is developed at length in the section below.

The conditions most frequently proposed for durable peace in Afghanistan, including inclusive governance, a unified national policy framework in place of parallel tribal authority structures, and protection of minority and women’s rights, are not free-standing normative goals in the framework developed here. Read through neoclassical realism, each condition addresses the internal-fragmentation variable specifically: an inclusive settlement is, in effect, an attempt to raise Afghan state cohesion to the point where it can withstand external pressure rather than being reshaped by it. The exclusion of women from the 2016 Hezb-i-Islami peace agreement, and the absence of any binding implementation mechanism for the rights it nominally recognized, illustrates the cost of treating such provisions as rhetorical rather than structural ([25]); durable inclusion requires women’s participation in agenda-setting stages of negotiation, not merely acknowledgment in a settlement’s final text.

Developments since 2021 offer limited grounds for optimism on this front. As of mid-2026, Russia remains the only state to have formally recognized the Islamic Emirate of Afghanistan, having done so in July 2025; other states, including China and several Central Asian neighbors, maintain a diplomatic presence in Kabul without extending formal recognition, a pattern of conditional, de facto engagement that international law scholarship describes as an unresolved and legally unsettled middle ground rather than a stable equilibrium ([24]; [9]; [24]). Domestically, the Taliban leadership has prioritized internal cohesion within its own movement over the inclusive governance most states have set as a condition for fuller recognition, consolidating a centralized, ideologically uniform administration even as this consolidation deepens its international isolation ([2]). Internally, the most recent UN human rights reporting documents a continued and, in several respects, intensifying restriction of women’s rights: no decree limiting women’s participation in public life has been reversed since 2021, and since September 2025, Taliban authorities have barred women, including UN national staff, from UN premises across the country ([4]; [25]). Human Rights Watch’s most recent country assessment documents continuing public floggings and executions through 2025 alongside a deepening humanitarian crisis, compounded by reduced international aid and the forced return of millions of Afghans from Iran and Pakistan ([10]). None of this indicates the kind of internal cohesion-building the framework developed here identifies as a precondition for durable peace.

The Afghanistan-Pakistan conflict introduced in Section 4.1 raises the sharper question of whether the five-stage pattern identified in this article is already producing a sixth stage. The documented sequence of events is as follows. Cross-border clashes escalated sharply in October 2025, following a surge in militant attacks inside Pakistan that Islamabad attributed to TTP fighters operating from Afghan soil, an accusation Kabul has consistently denied ([11]; [13]). A ceasefire followed, but tensions resumed in early 2026; by late February, Pakistani air and drone strikes had hit at least eight Afghan provinces, the highest number recorded in a single month since the Taliban’s 2021 takeover, and Afghan forces retaliated along the Durand Line, with fighting reaching Kabul and Kandahar ([11]). Displacement and disruption to cross-border trade have been substantial, though precise, independently verified figures remain difficult to establish this early in an active conflict; this article treats the widely reported order of magnitude, in the hundreds of thousands displaced, as provisional rather than settled ([11]).

What follows is interpretation rather than documented fact, and is offered as such. Reading this sequence as a contemporary instance of the mechanism traced across the preceding five stages requires treating the Taliban-TTP relationship as a case of unresolved internal control rather than simple deniability: the two groups share documented ideological and personal ties dating to the Afghan Taliban’s own formation, and Pakistani officials describe the TTP as receiving logistical and material support from the Taliban authorities, a characterization the Taliban dispute ([13]). On this reading, the proximate trigger, Pakistani accusations that the Taliban shelter the TTP and Taliban denial of the same, is an instance of exactly the mechanism this article has traced across two centuries: an externally drawn frontier (the 1893 Durand Line) intersecting with an internal Afghan actor whose relationship to the central government remains unresolved. Whether this develops into a sustained sixth conflict, or is contained through further international mediation of the kind that produced the October 2025 ceasefire, is not yet resolved as of this writing. The pattern by which it has emerged, however, is difficult to distinguish from the pattern this article has identified in the preceding five cases. It is worth stating explicitly what would count as evidence against this reading, rather than treating the sixth-stage interpretation as unfalsifiable. A negotiated resolution that produces durable separation between the Taliban government and the TTP without requiring sustained external mediation would suggest the Taliban possess exactly the internal cohesion and monopoly on organized violence that the actors in the preceding five stages lacked at equivalent moments. Such an outcome would accordingly count against, not for, the claim that this is a sixth iteration of the historical pattern. The available evidence as of early 2026 does not show this, but the possibility should be stated rather than assumed away.

These conclusions warrant an explicit caveat. Post-2021 developments postdate the bulk of the secondary source base this article and its underlying research draw upon, and the Afghanistan-Pakistan conflict in particular remains an active, rapidly evolving situation as of mid-2026. The claim advanced here, that this conflict fits the historical pattern identified in Sections 3 and 4, is offered as a plausible extension of the framework rather than a settled finding, and is a natural focus for future research once more complete documentation of the conflict’s causes and trajectory becomes available.

This article has argued that Afghanistan’s five major wars since 1839 are not best understood as isolated historical episodes but as stages in a single causal sequence, in which each conflict left external dependency and internal fragmentation unresolved, creating the conditions for the next external intervention. Neoclassical realism, which treats domestic state weakness as an intervening variable that filters and shapes the effects of systemic pressure, explains this pattern more fully than either a purely structural account of great-power competition or a purely internalist account of Afghan governance failure could on its own. The article’s central contribution is to show that external leverage and internal fragmentation are not additive but interacting causes: external powers have repeatedly exploited pre-existing Afghan fragmentation, while internal Afghan actors have repeatedly sought external patronage to prevail in domestic contests, with each dynamic reinforcing the other across two centuries.

Before extending this argument into a feedback-loop claim, it is necessary to distinguish two things that the analysis in Section 4.2 has so far treated together: the pre-existing decentralized structure of Afghan governance, and the additional fragmentation that repeated external intervention has produced on top of it. The former is a structural feature of Afghan political organization rather than a policy failure. A mountainous, multi-ethnic territory historically governed through negotiated relationships between a central court and largely autonomous local and tribal authorities is not thereby weak; it is organized differently from a centralized bureaucratic state, in a manner that required no strong central monopoly on violence until external powers made one strategically necessary ([1]). What repeated external intervention has done, from the nineteenth century onward, is convert this pre-existing plurality into something more brittle and more conflict-prone: arming particular factions over others, elevating a single claimant to central authority while leaving rival power centers militarily intact, and attaching international legitimacy to whichever government an external patron happened to back. It is this second, intervention-produced layer, not the underlying decentralized structure itself, that the feedback-loop argument below refers to. Each external intervention examined in this article left behind not simply an unresolved distribution of local authority, which Afghanistan had long managed on its own terms, but an actively militarized and internationally contested one, which is a materially different and more conflict-prone condition than the one that preceded it.

This case also carries a broader theoretical implication for neoclassical realism itself. The theory typically treats domestic state weakness as a pre-existing condition that shapes how a state responds to a given instance of external pressure ([19]). The Afghan case suggests a further possibility worth developing in future theoretical work: that repeated external intervention can itself be a cause of the specific, intervention-produced fragmentation that then shapes the state’s response to the next intervention. This is a feedback loop between the systemic and domestic levels that the existing neoclassical realist literature has not fully theorized, and one that buffer states subject to repeated great-power attention may be especially prone to.

Three policy implications follow directly from this framework, rather than from a general survey of Afghanistan’s problems. First, international engagement with Afghanistan, whether through aid, diplomatic recognition, or security cooperation, should be conditioned on verifiable progress toward internal power-sharing mechanisms, not solely on counterterrorism cooperation; the historical pattern traced here suggests that aid decoupled from institution-building tends to reproduce the fragility it was meant to address. Second, durable peace requires negotiating directly with Afghanistan’s internal fragmentation. This means engaging the full range of factional, ethnic, and regional actors rather than treating whichever faction controls Kabul as equivalent to a cohesive Afghan state. Third, and specifically in light of the evidence reviewed in Section 5, durable settlements require the binding, early-stage inclusion of women in negotiation processes, not rhetorical commitments added after the terms of a settlement are already fixed ([25]).

This analysis has clear limitations. It relies on qualitative synthesis of secondary sources rather than original archival or field research, and it examines a single case, which constrains the generalizability of its theoretical claims beyond buffer states with a broadly similar structural position. Its treatment of the post-2021 period, and especially of the still-unfolding Afghanistan-Pakistan conflict, is necessarily provisional given how recent and unsettled these events remain. Future research might extend the comparative element of this framework to other buffer states with comparable histories of repeated external intervention, or might revisit the sixth-stage question raised here once the trajectory of the current Afghanistan-Pakistan conflict becomes clearer. What the historical record examined in this article does suggest, with reasonable confidence, is that peace in Afghanistan is unlikely to prove durable so long as external engagement and internal fragmentation continue to be addressed separately rather than as the interacting problem this article has shown them to be.