The 1993 Eritrean Referendum: Independence as Constitutional Departure

Thesis

The Eritrean independence referendum of 23–25 April 1993 is the single most consequential constitutional act in the modern history of Ethiopia. It dissolved a federation that had existed since 1952 and a state union that had existed since 1962, established a sovereign Eritrean state on Ethiopia’s northern border, and removed Ethiopia from direct access to the Red Sea — a strategic loss whose consequences continue to shape Ethiopian foreign policy as of mid-2026, including the unfolding Ethiopia-Eritrea tensions over Assab.¹ The referendum was conducted under the constitutional authority of the Transitional Period Charter (Article 2 of the foreign relations section), supervised by a UN observer mission, and produced a 99.81 percent affirmative vote on a reported turnout of 98.5 percent.² Eritrean independence took effect on 24 May 1993.

The article’s thesis is twofold. First, the referendum is the clearest case in Ethiopian constitutional history of a constitutional process actually delivering what its formal procedures promised: a vote was held, the result was respected, sovereignty changed hands peacefully. This is genuine and worth recognising. Second, the conditions under which this happened — military victory by the EPLF, EPRDF acquiescence in the result before the vote was held, the absence of any organised Ethiopian political force capable of contesting the question — were so specific to the moment that the referendum cannot serve as a precedent for any later Ethiopian self-determination question. The Eritrean precedent did not democratise the Ethiopian state. It removed a region from it. The constitutional lesson Ethiopian politicians drew from 1993 was not “self-determination questions can be resolved democratically.” It was “the constitutional right of secession, once invoked seriously, ends in the loss of territory.” That lesson has shaped every Ethiopian political actor’s relationship with Article 39 of the 1995 Constitution since.

The referendum

The referendum was the culmination of thirty years of armed struggle by the EPLF and its predecessors against Ethiopian rule. The EPLF had taken effective control of most of Eritrea by mid-1990, captured Asmara on 24 May 1991, and operated as the de facto government of the territory from that date.³ The London peace conference and the Transitional Period Charter both ratified the EPLF’s authority pending a referendum. The referendum was scheduled by agreement between the EPLF and the Ethiopian Transitional Government in early 1992, with UN supervision through UNOVER (the United Nations Observer Mission to Verify the Referendum in Eritrea).

The ballot question was unambiguous: “Do you want Eritrea to become an independent sovereign state? Yes / No.” Voting took place over three days, 23–25 April 1993. UNOVER monitored polling and certified the conduct as free and fair within the conditions that prevailed in EPLF-controlled Eritrea.⁴ The reported result — 99.81 percent yes — is consistent with the political reality the war had produced and was not seriously contested by any party. Ethiopia recognised the result; the international community recognised the new state; Eritrea joined the UN on 28 May 1993.

In strictly procedural terms, this is what a successful constitutional self-determination referendum looks like.

The constitutional structure

The referendum’s constitutional authority derived from the Transitional Period Charter’s Article 2 (foreign relations section), which committed the transitional government to “respect the right of self-determination, including independence, of the Eritrean people, to be exercised through a referendum to be conducted under international supervision.”⁵ This commitment, in turn, derived from the London understandings of May 1991, which themselves derived from the military fact of the EPLF’s victory.

Constitutionally, the referendum was not a 1995-Constitution Article 39 process — that article did not yet exist. It was a Transitional Period Charter process. The procedural framework was created specifically for the Eritrean question and applied only to it. The drafters of the 1995 Constitution, when they wrote Article 39’s secession provisions, were drafting against the background of the Eritrean precedent. The 1995 framers knew what self-determination could look like when actually exercised; they wrote Article 39 with that knowledge.

Article 39 of the 1995 Constitution is consequently not naive about secession. It establishes a multi-step process: a two-thirds vote by the nation/nationality/people’s regional legislative council requesting secession, federal government organisation of a referendum within three years, the referendum producing a majority in favour, and the federal government transferring power to the seceding state and overseeing division of assets and territory.⁶ The drafters knew that secession was possible, that under sufficiently extreme conditions it would happen, and that the right form of constitutional procedure was to make it difficult but not impossible.

Whether this procedural elaborateness has produced any actual constraint on secessionist movements is a different question, taken up in Article 10 of this series. For now, the relevant point is that the Eritrean referendum established the operational meaning of self-determination in Ethiopian constitutional practice: a regional movement with the military means to control its territory and the political will to demand independence can, given the right international and domestic configuration, obtain it through constitutional process. This is more than a theoretical proposition. It happened.

What the precedent did

The Eritrean precedent shaped Ethiopian constitutional politics for the next three decades in four specific ways.

First, it established that the right of secession could be exercised in fact, not just in text. Many constitutions contain secession provisions that are universally understood to be unusable. The Eritrean precedent meant that Article 39 of the 1995 Constitution was not in that category. Every regional government with secessionist sympathies — and at various moments this included parts of the OLF, parts of the ONLF in the Somali region, and the TPLF in its post-2018 break with the federal government — could point to 1993 and say: it has been done; it can be done again.

Second, the precedent confirmed for non-secessionist political actors that the constitutional right to secession was, in practice, a constitutional pathway to state dissolution. Defenders of Ethiopian unity from the 1990s onward have invoked 1993 not as a model but as a warning. The CUD platform in 2005, the various Amhara nationalist movements that emerged after 2018, and significant parts of the urban Ethiopian intelligentsia have argued that Article 39 must be removed from the constitution because its precedent — the Eritrean precedent — is what state dissolution looks like.⁷ The salience of the 1993 referendum in this argument is consistent and operational.

Third, the precedent made the question of the Eritrean port — first Massawa, then increasingly Assab — a permanent feature of Ethiopian strategic thinking. Ethiopia became landlocked at midnight on 24 May 1993. Every Ethiopian government since has had to manage the consequences. The 1998–2000 Eritrean-Ethiopian war was partly about Assab. The post-2018 deterioration of Ethiopian-Eritrean relations, after a brief 2018–2020 rapprochement, has Assab at its centre — Prime Minister Abiy Ahmed has identified Assab as a potential location for Ethiopian sea access, and Eritrea has treated the question as a threat to its sovereignty.⁸ As of mid-2026, the question is unresolved and is contributing to an increasingly probable renewed conflict.

Fourth, the precedent put the question of constitutional design under the shadow of irreversibility. Constitutional choices about federalism and self-determination, the Eritrean experience suggested, were not provisional architectural decisions that could be revisited. They were decisions whose consequences could be the permanent loss of territory. This made the politics of constitutional amendment harder than it would otherwise have been, and made every subsequent constitutional debate — about Article 39, about the structure of the federation, about the place of nationality in the constitutional order — heavier with implication.

The Badme war

The 1998–2000 Eritrean-Ethiopian war is not, strictly, the subject of this article — it is taken up in the Birr series treatment of the monetary divorce. But the war is part of the constitutional story because it is what happened when the Eritrean precedent’s basic premise — that independence could be peacefully decoupled from continued bilateral cooperation — failed. The war began as a border skirmish in Badme on 6 May 1998, escalated within weeks to full conventional war, killed between 70,000 and 100,000 soldiers and civilians, and ended with the Algiers Peace Agreement of 12 December 2000 and the Ethio-Eritrea Boundary Commission ruling of April 2002, which Ethiopia did not implement.⁹ The unresolved border dispute, the disrupted economic relationship, and the personal hostility between the two ruling parties (TPLF in Ethiopia, EPLF/PFDJ in Eritrea) shaped the next eighteen years until Abiy Ahmed’s 2018 rapprochement, and have shaped the renewed deterioration since 2022.

The constitutional lesson of Badme, as opposed to the lesson of 1993, is that even a successful constitutional self-determination process can fail to produce the political stability it promised. The form of the 1993 referendum was a constitutional success. The substantive consequence — a stable settlement between the two states — was a constitutional failure. The two evaluations are not contradictory; they operate at different levels of analysis. The procedural success of 1993 does not redeem the substantive failure of 1998–2002, and vice versa.

Steelman of the opposing reading

Three opposing readings of the 1993 referendum and its constitutional significance deserve engagement.

The first is the EPLF/Eritrean reading: that the referendum was the inevitable and just consummation of a thirty-year struggle for Eritrean self-determination, that the Ethiopian state had no legitimate claim to retain Eritrea after the 1962 dissolution of the federation, and that the constitutional choreography of 1991–93 simply formalised what the war had already decided.¹⁰ This is largely correct as a historical reading. Ethiopian rule over Eritrea after 1962 was not legitimate by any meaningful standard, and the EPLF’s victory by 1991 was a fact that constitutional process could only acknowledge, not reverse. The argument of this article is not that the referendum produced an unjust outcome. The argument is that it produced a precedent whose constitutional consequences for the rump Ethiopian state were profound and largely negative.

The second is the unionist Ethiopian reading: that the referendum was a strategic disaster for Ethiopia — the loss of sea access, the creation of a hostile northern neighbour, the elevation of TPLF leaders who had collaborated with the EPLF to power in Addis Ababa — and that no constitutional process should have ratified it.¹¹ This reading has gained force since 2018 and is the foundation of the contemporary Ethiopian nationalist critique of the EPRDF settlement. The reply is that constitutional process did not produce the strategic outcome; military fact did. The EPLF had won the war by 1991. The constitutional choice was whether to ratify that fact peacefully or refuse to ratify it and produce a continuing conflict. The peaceful ratification was, in the conditions of 1991, the better available outcome. The post-1993 deterioration is real but its root cause is the unresolved relationship between the two states, not the act of ratification.

The third is the federalist Ethiopian reading: that the Eritrean referendum was a one-off case, justified by the specific history of the 1952–62 federation and the thirty-year war, and that it does not constitute a precedent for any Ethiopian region’s secession because no other region has comparable history.¹² This is partly right and partly wrong. It is partly right that the Eritrean case is sui generis in important respects. It is partly wrong because the constitutional language of Article 39 does not distinguish between cases with Eritrean-style histories and cases without; it grants the same procedural rights to every nation/nationality/people in Ethiopia. The fact that no other region has yet exercised the right does not mean none will. The recent Tigray war and the post-2018 collapse of the Eritrean-Ethiopian rapprochement both put the Article 39 question back in operational play.

Multi-causal accounting

The 1993 outcome reflects (a) the EPLF’s military victory in 1990–91, the proximate cause; (b) the international diplomatic context of the early 1990s, in which the post-Cold War US position favoured peaceful resolution of African conflicts through self-determination mechanisms; (c) the willingness of the EPRDF (and particularly the TPLF, which had operated as the EPLF’s junior partner during the late war) to accept Eritrean independence as part of the political settlement; (d) the absence of any organised Ethiopian political force capable of contesting the settlement in 1991–93; and (e) the specific history of the 1952–62 federation and its forced dissolution. The constitutional process did not produce the outcome; it ratified an outcome produced by other forces. The constitutional significance is in the ratification, which set the framework subsequent Ethiopian constitutional politics has operated within.

What the article establishes

The 1993 Eritrean referendum is the constitutional act that established, in Ethiopian practice, the operational reality of self-determination as state dissolution. It was procedurally successful by any reasonable standard. It produced consequences for the rump Ethiopian state — landlocked status, a hostile northern border, a permanent question about the legitimacy of Article 39’s secession provisions — that continue to shape Ethiopian constitutional politics as of mid-2026, including the most acute current crisis in Tigray and the unresolved Assab question. The referendum’s procedural success and its substantive consequences cannot be separated; the constitutional record contains both.

The next article takes up the document that the Transitional Period Charter was supposed to produce: the 1995 Constitution of the Federal Democratic Republic of Ethiopia, and its theory of ethnic federalism as a theory of the state.

---

References

Footnotes

1. On the strategic consequences of Ethiopian landlocked status, see Patrick Gilkes and Martin Plaut, War in the Horn: The Conflict between Eritrea and Ethiopia (London: RIIA, 1999); for the contemporary Assab question, “The growing threat of conflict in the Horn of Africa,” The New Humanitarian, March 2026, https://www.thenewhumanitarian.org/analysis/2026/02/23/growing-threat-conflict-horn-africa. ↩

2. UNOVER, Report of the United Nations Observer Mission to Verify the Referendum in Eritrea (New York: UN, May 1993). Eritrean independence took effect 24 May 1993; UN membership 28 May 1993. ↩

3. Ruth Iyob, The Eritrean Struggle for Independence (Cambridge: Cambridge University Press, 1995), chs. 6–8; Dan Connell, Against All Odds: A Chronicle of the Eritrean Revolution (Trenton: Red Sea Press, 1997). ↩

4. UNOVER report and the contemporary observer accounts. ↩

5. Transitional Period Charter, Article 2 (foreign relations section). ↩

6. 1995 FDRE Constitution, Article 39. ↩

7. On the contemporary critique, see Tronvoll and Berhe, “The Slow Death of the Ethiopian Federation,” in Ethiopia: Constitutional Crisis and the New Federalism, ed. K. Tronvoll (forthcoming); and “The Faultlines Of A Nation,” The Reporter Ethiopia, November 2025, https://www.thereporterethiopia.com/47667. ↩

8. For the contemporary Eritrea-Ethiopia situation, see “Military Buildup In Northern Ethiopia,” Africa File, February 12, 2026, https://www.criticalthreats.org/analysis/tplf-eritrea-ethiopia-fardc-m23-drc-splmio-south-sudan-boko-haram-africom-africa-file-february-12-2026. ↩

9. On the Badme war, see Tekeste Negash and Kjetil Tronvoll, Brothers at War: Making Sense of the Eritrean-Ethiopian War (Oxford: James Currey, 2000); Martin Plaut, Understanding Eritrea: Inside Africa’s Most Repressive State (London: Hurst, 2016). ↩

10. This is broadly the position of Eritrean nationalist historiography, e.g., Connell, Against All Odds; Iyob, The Eritrean Struggle. ↩

11. This is the position of the contemporary Ethiopian nationalist critique. See the discussion in Aregawi Berhe, A Political History of the Tigray People’s Liberation Front (Los Angeles: Tsehai, 2009). ↩

12. This is the position taken by EPRDF-aligned scholars in the 1990s and early 2000s, e.g., Andreas Eshete and Fasil Nahum. ↩