Return on Whose Terms? What The Gambia v. Myanmar Should Say About Going Home

[Kazi Ayman Awsaf is an undergraduate student and teaching assistant at East West University (Bangladesh).]

On 29 September, the first 1,500 of the 5,000 Rohingya whom Myanmar has agreed to take back from Malaysia are due to leave, in what Kuala Lumpur calls a voluntary programme, aboard two Myanmar navy warships and a hospital ship. Human Rights Watch has warned that “conditions for safe, dignified, and voluntary return to Myanmar do not currently exist.” In The Hague, the International Court of Justice has been deliberating since the merits hearings closed on 29 January on The Gambia’s claim that Myanmar committed genocide against the Rohingya. Observers did not expect a judgment before the end of the year.

Commentary on the pending judgment has mostly concerned liability: the threshold for inferring genocidal intent and sexual and reproductive violence as genocidal acts. Khan Khalid Adnan has rightly called remedies “the real test” of the case. This post narrows that test to one remedy and to the place where it will be lived. Since its Application (para. 112), The Gambia has asked the Court to order, as reparation for the victims, the “safe and dignified return” of displaced Rohingya, a request it repeated in its final submissions (CR 2026/20, p. 58). Around 1.2 million Rohingya now live in the camps of Cox’s Bazar, Bangladesh, some 150,000 of them arrivals since mid-2024.

I argue that what the Court says about return will travel further than anything it says about intent, and that it should say three things. First, reparation claimed by a non-injured State is owed in the interest of the beneficiaries, so return must be framed as a right the Rohingya may exercise, never an obligation that others may invoke against them. Second, restitution of legal status must come before physical return, not after it. Third, Myanmar’s loss of control over northern Rakhine does not dissolve its obligation; it changes what the obligation requires today.

The Declaratory-Only Defence and the Beneficiary Principle

Myanmar asks the Court to reject all of The Gambia’s remedial claims (CR 2026/23, p. 31, para. 6). As Michael Becker summarised its position, a non-injured applicant is “limited to declaratory relief” and cannot obtain restitution or compensation. Writing on this blog in 2019, Dimitrios Kourtis observed that the ILC’s commentary, citing Namibia, “seemingly support[s]” that view, and thought anything more “unlikely, given the questions of causation, valuation, and identification.”

The Court should reject the declaratory-only reading. Article 48(2)(b) of the ILC Articles on State Responsibility allows any State entitled to invoke a collective obligation to claim “performance of the obligation of reparation … in the interest of the injured State or of the beneficiaries of the obligation breached.” The ILC conceded that this “involves a measure of progressive development,” but one “justified since it provides a means of protecting the community or collective interest at stake” (Commentary to Art. 48, para. 12).

The Court’s reasoning on standing points the same way. In 2022 it held that all parties share a “common interest to ensure the prevention, suppression and punishment of genocide” (paras. 106–114); in Belgium v. Senegal it had reasoned that if a special interest were required, “in many cases no State would be in the position to make such a claim” (para. 69). The same is true of reparation. When a State is accused of destroying part of its own population, the injured have no State to claim for them. A declaratory-only rule would give the Convention a court that can find genocide but cannot order it repaired.

Kourtis’s three obstacles are also weaker here than they looked in 2019. Causation is a problem for failures to prevent, not for commission. Valuation can be met with global sums. And identification is unusually tractable: most beneficiaries live in one district of one neighbouring State.

The commentary, however, also sets the principle’s limit. A State claiming “anything more than a declaratory remedy and cessation may be called on to establish that it is acting in the interest of the injured party” (para. 12). Here that interest belongs to the Rohingya, and they have stated it. As the camps marked the ninth anniversary of the 2017 exodus, Nay San Lwin of the Free Rohingya Coalition summarised the demand: to “return to our homeland with full rights, peace, safety, dignity and justice.” A return that its beneficiaries cannot refuse is not reparation in their interest. It is another displacement, run in reverse.

Why the Operative Paragraph Will Be Read in Dhaka and Kuala Lumpur

A judgment binds only the parties (Article 59 of the Statute), and Bangladesh is not one of them. It acceded to the Convention in 1998 with a reservation requiring “the consent of all parties to the dispute” before any case goes to the Court under Article IX. The Court held in 2022 that this reservation does not affect The Gambia’s standing (Pillai), but it would have complicated any intervention: in Ukraine v. Russia the Court held that a State with an Article IX reservation could not intervene on the construction of that provision, nor, at that stage, of the Convention’s other provisions (Order of 5 June 2023). The State hosting the largest share of the beneficiaries has had no voice in designing their remedy.

Yet the remedy’s wording will be read most closely in Dhaka. The Bangladesh Nationalist Party, which won February’s election, calls repatriation “a matter of utmost priority” and points to the repatriation of 1992, under Khaleda Zia’s government, as a success. That history should give the Court pause. After the 1978 exodus, returns rose only once Bangladesh “allowed camp conditions to decline and restricted food rations”; in 1992, UNHCR withdrew its support “when it became clear that coercion was continuing“, yet some 230,000 refugees went back between 1993 and 1997. They returned to the statelessness that the 1982 Citizenship Law had formalised, and to the conditions that preceded the violence of 2012 and the expulsions of 2016 and 2017.

None of this singles out Dhaka. Bangladesh has carried for nine years a burden that wealthier States decline to share, as the 26 percent cut to the 2026 Joint Response Plan shows. But it is bound by non-refoulement under Article 3 of the Convention against Torture and, as the Human Rights Committee reads Articles 6 and 7, the ICCPR (General Comment No. 31, para. 12). It also designates the Rohingya “Forcibly Displaced Myanmar Nationals” rather than refugees. A judgment ordering Myanmar to “allow” return, without more, invites the reading that return is a matter between governments. Malaysia’s Prime Minister has already framed it that way: “Myanmar must accept them back.”

The Court should therefore state, in the operative clause or in reasoning tied closely to it, that return must be voluntary, informed, safe, dignified and sustainable, and that nothing in the judgment implies those conditions exist today. The Court cannot bind Bangladesh or Malaysia. It can stop its judgment being used against the people it is meant to repair.

Status Before Return

The Gambia frames “full and equal citizenship” as a guarantee of non-repetition (CR 2026/20, p. 59). It is also the first form of restitution, and the precondition for the rest. The Court’s practice allows it to impose an obligation of result while leaving the means to the respondent: “review and reconsideration” by “means of its own choosing” in Avena (para. 153(9)); “the choice of means must be left to the United States” in LaGrand (para. 125). The Court need not redraft the 1982 law. It can require that no Rohingya be denied nationality or identity documents on discriminatory grounds, and that the right to identify as Rohingya, which The Gambia also claims (p. 58), be respected. And it can make explicit that physical return is not the measure of compliance.

The sequence matters because every earlier return ran the other way: people first, rights promised later, and in the end no rights at all.

Control Lost, Obligation Retained

The hardest fact for the remedy is that Myanmar no longer governs the places people would return to. Since the Arakan Army captured Maungdaw in December 2024, the entire frontier with Bangladesh has run through territory it administers. Human Rights Watch has documented an Arakan Army massacre of Rohingya villagers in Buthidaung in May 2024 and reports that survivors remain effectively detained. The respondent State, meanwhile, is led by Min Aung Hlaing, president since April, for whom the ICC Prosecutor has sought an arrest warrant for the deportation and persecution of the Rohingya.

Loss of control does not extinguish responsibility. Restitution is owed unless, and only “to the extent” that, it is “materially impossible” (ARSIWA, Art. 35(a)). What changes is the content of the obligation. Restoring homes and communal sites in Arakan Army–held townships (CR 2026/20, p. 58) may be materially impossible for now, and compensation must cover the gap (Art. 36). DRC v. Uganda shows the Court can award a global sum for mass harm, there US$325 million (para. 106), though the method has been criticised for obscuring what individual victims receive. That problem is sharper here: the DRC could pass the money to its own nationals, while The Gambia has no such link to the beneficiaries. Any award must go to a mechanism designed with Rohingya participation and reachable from Cox’s Bazar. Legal restitution, by contrast, needs no territorial control at all. Citizenship is legislated in Naypyidaw, not in Maungdaw.

Rather than fix a return remedy now for conditions that will change, the Court should reserve the modalities of restitution and compensation for a later phase, as it did in Armed Activities in 2005 (para. 345(6)). It already has a monitoring template. Under paragraph 86(4) of the 2020 provisional measures order, Myanmar had filed twelve reports by November 2025. That duty lapses with the final decision; a reserved reparation phase would keep both the case and the scrutiny alive.

Two Objections

The first objection is that the Court should not address States that are not before it. It need not. It needs only to define what Myanmar’s obligation of return consists of, and voluntariness is part of that content, not an instruction to Dhaka. The second is that stressing voluntariness hands Myanmar an excuse for inaction. It does the opposite when paired with obligations of result on legal status that do not depend on anyone’s return. Myanmar can comply with those tomorrow. It cannot shelter behind refugees’ refusal to return to territory it does not control.

Conclusion

The Court has found a breach of the Convention only once. In Bosnia it treated a declaration as sufficient satisfaction for Serbia’s failure to prevent genocide, because it had not been shown that the genocide would have been averted (paras. 462–463). If it finds that Myanmar itself committed genocide, that causal obstacle falls away, and the Court will have to say for the first time what reparation for genocide by a State against its own people looks like. In Cox’s Bazar the judgment will face a simpler test: does it give the Rohingya a right to go home on their own terms, or give others a reason to send them?

The author has no connection with the parties, the intervening States or counsel in The Gambia v. Myanmar, and no governmental, institutional or organisational affiliation relevant to this post.

Photo attribution: Photo by Rohingya Creative Production on Unsplash