Digital resources in the Social Sciences and Humanities OpenEdition Our platforms OpenEdition Books OpenEdition Journals Hypotheses Calenda Libraries OpenEdition Freemium Follow us

#8 An electoral system for today’s Brussels

Céline Romainville (UCLouvain), Dave Sinardet (VUB), Émilie Van Haute (ULB), Guillaume Delvaux (UCLouvain), Kris Deschouwer (VUB), Petra Meier (UAntwerpen), Philippe Van Parijs (UCLouvain), Willem Sas (UHasselt)

The image shows a blue ballot box on an orange background. A ballot paper with two columns is being inserted into it.
Illustration: Brussels Studies, 2025.

Translation: Jane Corrigan


Un système électoral pour le Bruxelles d’aujourd’hui

Een kiessysteem voor het Brussel van vandaag


The BSI Position Papers reflect the opinions, thoughts and recommendations of the authors, and are not attributable in any way to Brussels Studies and the Brussels Studies Institute (BSI). The authors assume full responsibility for them.


Authors’ note: This text is the result of a collective reflection carried out as part of the Re-Bel  initiative (Rethinking Belgium’s Institutions). It is informed by two workshops involving academics and Brussels politicians (University Foundation, 20 February and 27 March 2025), various meetings with other Brussels politicians, and the valuable feedback of many colleagues. The argument presented here in summary form is developed in greater depth in a Re-Bel e-book shaped by these exchanges.


The Brussels electoral system is based on an institutional structure designed for a sociolinguistic configuration which no longer exists. Faced with an increasingly diverse population and political issues which go beyond traditional linguistic divides, maintaining the dual electoral college now appears to hinder representativeness and democratic effectiveness. This text puts forward a reform model which guarantees the representation of linguistic groups without compartmentalising the electorate.

A system designed for a Brussels which no longer exists

The system which currently governs Brussels elections was designed for a Brussels which no longer exists. When it was conceived and negotiated in the 1970s and 1980s, foreigners and Belgians of recent foreign origin, as defined by Statbel (citizens with at least one parent born a non-Belgian), made up only a modest share of the city’s population. The linguistic question was central to Brussels politics, and it was difficult to imagine a French speaker voting for a Dutch speaker or vice versa. In that context, a system composed of two electoral colleges allowed the Dutch-speaking minority to choose its representatives separately for the regional parliament and government, thus defending its legitimate interests.

Today, Belgians with two Belgian-born parents account for just 22 % of the population of Brussels1. Only 40 % of Brussels residents (including 6 % of French nationals) have French as their only mother tongue, and just 5 % have Dutch as their only mother tongue2. The most emotionally charged political issues in Brussels now relate less to language than to governance, security, mobility or secularism. All parties in the Dutch-speaking college campaign in both languages. According to the most reliable estimates, more than 40 % of their voters are more French-speaking than Dutch-speaking3. At both federal and municipal levels, the vast majority of political families now present joint lists.

Following the June 2024 elections, the problems inherent in the current electoral system are plain to see. This did not prevent the election within the Dutch-speaking college of a candidate unable to express himself in Dutch. It has also created fertile ground for the multiplication of vetoes which have paralysed government formation for months.

Guaranteed representation without a double college

A reform of this electoral system is necessary. Those aspiring to become members of parliament or even ministers must be able to seek support from the entirety of the Brussels electorate – not only from a linguistically defined fragment. The government emerging from elections must no longer be the forced product of two independently formed majorities. Bilingual lists should be permitted, as they are in federal and municipal elections. Yet the guaranteed representation of Dutch speakers within the Brussels parliament and government must not be abolished.

This guaranteed representation remains essential to enable the formation of the assembly and executive of the Flemish Community Commission, to ensure double-majority votes on linguistically sensitive matters, to serve as a counterpart to the parity required within the federal government, and above all, to provide Brussels with two-way “ambassadors” capable of facilitating communication and cooperation with the surrounding region and a federal state with a Dutch-speaking majority.

In principle, guaranteed representation could be ensured by a formula guaranteeing that each language group has not a precise number of representatives (e.g. 72 French speakers and 17 Dutch speakers, as today), but a minimum number (e.g. at least 40 French speakers and at least 10 Dutch speakers). This would avoid requiring every candidate to identify formally as either French-speaking or Dutch-speaking. It would also have the advantage of not locking in for an indefinite period a ratio chosen more or less arbitrarily at a given time. Yet it would have the major drawback of jeopardising the inter-community peace achieved partly through neutralising the impact of each group’s electoral weight on the balance of power within parliament.

Therfore, for good reasons, guaranteed representation of both linguistic groups in the regional parliament should continue through a fixed number of seats. Two alternative formulas could achieve this within a single electoral college allowing both bilingual and unilingual lists. Each is compatible with the current 72/17 distribution as well as with any future modification of that ratio or the total number of seats. Both maintain the familiar framework of proportional representation, with seats allocated to lists according to the D’Hondt rule4, and allow voters to cast either list votes or preference votes for any number of candidates on the same list. In both formulas, candidates on bilingual lists may appear in two sub-lists – one Dutch-speaking and the other French-speaking. As with lists of effective and successor candidates for federal elections, each voter may vote for any number of candidates on both sub-lists, whether French or Dutch speakers.

As at present, each formula may require candidates on each of these sub-lists to hold an identity card in the relevant language and to be presented by an appropriate number of outgoing MPs or citizens. It would be unconstitutional to impose a formal language condition. However, as there would no longer be a Dutch-speaking college in which one could hope to obtain seats at lower cost5, one could reasonably expect the parties to put forward for each sub-list linguistically qualified candidates to sit on the assemblies of the corresponding Community Commissions (COCOF and VGC).

Parallel and corrective formulas

In the parallel formula, the 72 French-speaking and 17 Dutch-speaking seats are distributed according to the D’Hondt rule between the French-speaking and Dutch-speaking sub-lists, in proportion to the total number of votes received by all of the lists as a whole6. The seats allocated to each sub-list are then assigned to candidates according to their personal votes and, if applicable, (common) list votes.

In the corrective formula, all 89 seats are distributed proportionally among the lists according to the number of votes obtained, regardless of linguistic group, and then initially assigned to candidates, whether French- or Dutch-speaking, based on their personal votes and, if applicable, the list votes which they could benefit from. If the resulting distribution deviates from the 72/17 quotas, the seat(s) exceeding the quota guaranteed to one language group are allocated to the first eligible candidate on the other sub-list of the same list if it is bilingual. If the list is unilingual, surplus seats are reassigned to the next eligible candidate on the following list, according to the D’Hondt rule, provided it includes a sub-list of the required language group7.

The most significant difference between the two formulas concerns parties wishing to present a unilingual list. Simulations show that under the parallel formula, a list (bilingual or unilingual) must secure close to 5 % of the total vote to gain at least one Dutch-speaking seat via the D’Hondt rule8. However, in June 2024, the leading Dutch-speaking party (Groen) obtained just 3.9 % of the total vote. Even if the current 5 % electoral threshold was abolished, the parallel formula would in effect render any unilingual Dutch-speaking list unviable.

Under the corrective formula, these simulations show that the D’Hondt rule can already award a first seat to a list with just under 1 % of the vote. And there is little chance that the Dutch-speaking quota would be filled before the allocation of a first seat to a unilingual Dutch-speaking list. The corrective formula, as such, is therefore compatible with the viability of unilingual Dutch-speaking lists. But maintaining the current 5 % electoral threshold is not. Lowering it, perhaps to 3 %, in order to give such lists a chance, could be justified as a specific requirement of Brussels democracy.

However, a more significant reduction – or elimination – of this electoral threshold would be undesirable, as it would cause excessive fragmentation within the parliament. Moreover, as the Vlaams Belang first demonstrated in the 1990s and the N-VA confirmed in Wallonia during the 2024 federal elections, explicitly Dutch-speaking parties can successfully attract non-Dutch-speaking voters.

Unlike the parallel formula, the corrective formula allows a variant in which bilingual lists do not appear as two sub-lists. Candidates remain identified as “Dutch-speaking” or “French-speaking”, but appear in any order on a single list. The advantage of this approach is that it contributes less than those with two sub-lists to perpetuating an image of the Brussels population which is disconnected from its current reality. This variant of the corrective formula requires agreement between French and Dutch speakers as to the order of candidates on the list. To preserve the autonomy of the Dutch-speaking component, it may include rules permitting the merging of lists independently composed by Dutch-speaking and French-speaking parties and guaranteeing the preservation of the ordering of candidates on the merged list. It may also impose the presence of candidates from each component in the top positions on each list.

Community parliaments

With or without sub-lists, the corrective formula – alongside a reduced electoral threshold – offers the most promising way to ensure a fixed representation of both linguistic groups without maintaining two distinct electoral colleges. Candidates elected under this system would form the assemblies of the community commissions (COCOF and VGC), as candidates elected in the two separate colleges do today. And Brussels members of the French Community parliament would continue to be elected from among the French-speaking members of the Brussels parliament.

However, a similar approach could not be used to elect Brussels members of the Flemish parliament, as this would impose an excessive workload on too few members. In use until 2004, it was replaced by a direct election by voters opting for the Dutch-speaking college.

As this system is no longer available, it could be replaced by a direct election of the six members of the Flemish parliament by Brussels voters holding a Dutch-language identity card. Due to the widespread use of electronic voting, this system does not pose any technical problems. It is at least as effective as the current system for identifying voters with a legitimate interest in the competences of the Flemish Community. And given the free and reversible choice of language for the identity card, it does not seem that it should be read as the creation of a “sub-nationality” in a sense strong enough to be considered problematic.

Double majority

The members of the Brussels parliament elected under this procedure will have been elected – most on bilingual lists – not by an “ethnically” defined subset, but by the entire Brussels regional electorate, to whom they would now be accountable. There would no longer be cheaper or costlier seats in terms of votes depending on the electoral college in which they are allocated. As a result, there would no longer be an opportunistic incentive to present lists and candidates in the college where seats are cheapest, nor any understandable challenge to the legitimacy of the parliamentarians and ministers elected in that college.

Within the Brussels parliament, both linguistic groups would retain particular responsibilities for votes on linguistically sensitive issues requiring a double majority. This double majority would also continue to be needed for the formation of the Brussels government for as long as its Dutch-speaking and French-speaking members form the executives of the VGC and COCOF. For them to operate effectively, they need a majority within their respective assemblies.However, it is reasonable to hope that this double majority requirement would not lead to the kind of deadlock which followed the June 2024 elections. Negotiations to form a regional government would occur among representatives of lists which would be largely bilingual, rather than separately between two sets of unilingual lists. Linguistic compromises, should they prove necessary, would be reached before government formation negotiations within most lists, and these negotiations would therefore be able to start more quickly on the many regional issues which have nothing to do with linguistic sensitivities.

In many respects, the situation in the Brussels region is alarming. Its capacity to respond must not be hampered by political institutions designed for a Brussels which no longer exists. It is high time for a serious discussion of concrete proposals which could make these institutions more effective and legitimate. Ours may not be the best. It does not claim to provide a cure for all possible deadlocks. It is, however, an open invitation to all those who believe they can offer better alternatives to articulate them clearly and demonstrate how their proposals would overcome the various challenges inevitably raised by reforms of this kind.


Auteurs

Guillaume Delvaux is a doctoral student in constitutional law at UCLouvain. His thesis focuses on the institutional future of the Brussels Region.

Kris Deschouwer is professor emeritus of political science at Vrije Universiteit Brussel. He has conducted research on political parties, elections and political representation. He has published several works, including The Politics of Belgium. Governing a Divided Society (2014).

Petra Meier is professor of political science at University of Antwerp. Her research focuses on the democratic character of the Belgian political system, and more specifically on the relationship between federalism and democracy, as well as on the participation and representation of demos and demoi. She is one of the editors of the Oxford Handbook on Belgian Politics (O.U.P, 2026) and one of the authors of The Politics of Belgium (3rd edition, Palgrave 2026).

Céline Romainville teaches constitutional law, the law of rights and freedoms and general state theory. Her research focuses on federalism and Brussels institutions, the parliamentary system and trust in political institutions, and constitutional law in the face of climate change.  She co-edited Quel fédéralisme pour la Belgique de demain (Larcier, 2024).

Willem Sas is professor of public finance at Hasselt University and research associate in the economics department at KU Leuven and the Center for Applied Public Economics at UCLouvain. His research interests include fiscal federalism, the budgetary organisation of the Belgian state, transfers between regions, the political impact of economic shocks and socio-political attitudes.

Dave Sinardet is Professor of Political Science at the Vrije Universiteit Brussel and UCLouvain Saint-Louis Bruxelles. His research on nationalism, federalism and multilingual democracy has appeared in journals such as West-European politics, GovernanceParty PoliticsGovernment and OppositionPublius: The Journal of Federalism en Public Opinion Quarterly. Since two decades, he has actively contributed his expertise to the public debate on the reform of Belgium’s institutions, with particular attention to the democratic and governance challenges in Brussels.

Emilie Van Haute is a professor in the Department of Political Science at Université libre de Bruxelles (ULB). She conducts research on political parties and electoral behaviour at the Centre d’étude de la vie politique (Cevipol). She is co-author of the Oxford Handbook of Belgian Politics (O.U.P. 2026) and The Politics of Belgium (3rd edition, Palgrave 2026).

Philippe Van Parijs is professor emeritus at UCLouvain (Hoover Chair of Economic and Social Ethics) and visiting professor at KU Leuven. With Paul De Grauwe, he founded the Re-Bel (Rethinking Belgium’s Institutions) initiative in 2009. Since 2020, he has chaired the Brussels Council for Multilingualism. He is the author of Linguistic Justice for Europe and for the World (OUP, 2011) and of Belgium. Une utopie pour notre temps/Een utopie voor onze tijd (Académie royale 2018).

Haut de la page ­^


To cite this post: ROMAINVILLE Céline, SINARDET Dave, VAN HAUTE Émilie, DELVAUX Guillaume, DESCHOUWER Kris, MEIER Petra, VAN PARIJS Philippe et SAS Willem, 2025. An electoral system for today’s Brussels. In: BSI Position Papers, no. 8, 27/11/2025. Available from: https://bsiposition.hypotheses.org/3827

  1. On 1 January 2025, according to figures published by Statbel, “Origin by nationality group detailed by province”. Available from: https://statbel.fgov.be/fr/themes/population/structure-de-la-population/origine#figures. []
  2. SAEYS Mathis, 2024. Taalbarometer 5. Factsheet. Brussels: Government of Flanders. Available from: https://www.briobrussel.be/node/19094 – survey end of 2023/beginning of 2024 []
  3. VAN HAUTE Émilie and BIESEMANS Romain, 2025. Vote, langues, et priorités politiques des Bruxelloises et Bruxellois. Brussels: Université libre de Bruxelles, CEVIPOL. Available from: https://notlikeus.be/wp-content/uploads/2024/05/BXL-NLU-1.pdf []
  4. The D’Hondt rule, used in Belgium for federal and regional (not municipal) elections, consists in assigning quotients to each list, obtained by dividing the number of votes cast for it by 1, 2, 3, 4, etc., then distributing the seats between the lists in the order of these quotients. For example, if four seats are to be shared between two lists having obtained 12,000 and 6,000 votes respectively, three seats will go to the first (quotients 12,000, 6,000, 4,000, 3,000) and one to the second (quotients 6,000, 3,000, 2,000, 1,500). []
  5. In 2009, the Dutch-speaking vote was at its lowest, with 11.2 % of the votes and a guaranteed 19.1 % of the seats (17 out of 89). A Dutch-speaking seat required an average of 3,048 votes and a French-speaking seat, 5,678. Since then, the gap has narrowed (17.1 % of the vote in the Dutch-speaking college in 2024), precisely in part due to strategic considerations on the part of a growing number of voters and because “linguistically asexual” lists opted for the Dutch-speaking college (Agora, Be.One, Volt, Viva Palestina, and above all Team Fouad Ahidar). []
  6. A list having obtained 10 % of the votes may therefore claim around 10 % of the 17 Dutch-speaking seats, provided that it has a Dutch-speaking sub-list, and around 10 % of the 72 French-speaking seats if it has a French-speaking sub-list. If the list is unilingual, it would have to settle with 10 % of the seats in its language group. []
  7. If, for example, the 89th seat allocated under the D’Hondt rule is the 73rd allocated to a French speaker, this seat would have to be transferred to the first non-elected Dutch-speaking candidate on the same list. If the list is unilingual, the seat would be transferred to the first non-elected Dutch-speaking candidate on the list (bilingual or unilingual Dutch-speaking) who would have won the 90th seat if there had been 90 seats in parliament. []
  8. These simulations, carried out by Kris Deschouwer, are based on the distribution of votes between bilingual and unilingual lists in the Brussels-Capital constituency in the June 2024 federal elections. Nothing, of course, guarantees that this distribution would be roughly the same in regional elections with a single college. However, simulations of this type make it possible to determine more realistically the minimum requirement for obtaining a seat than simply dividing 100 by 17 (5.9 %). Simulations available in: VAN PARIJS Philippe, 2025. Towards a new electoral system for the Brussels Capital Region, annexe 3. Available form : https://rethinkingbelgium.eu/ebook/toward-a-new-electoral-system-for-the-brussels-capital-region/ []

OpenEdition suggests that you cite this post as follows:
bsipositionpapers (November 27, 2025). #8 An electoral system for today’s Brussels. BSI Position Papers. Retrieved February 17, 2026 from https://doi.org/10.58079/158ax