Switzerland’s direct democracy: initiatives, referendums, the double majority and the Landsgemeinde

Swiss direct democracy is the set of federal procedures under which 100,000 eligible voters can force a national vote on a constitutional amendment, and 50,000 can force a vote on a law Parliament has passed.⁠¹ Voters are called to the polls about four times a year on an average of fifteen issues, and constitutional changes need a majority of voters and a majority of cantons.⁠^1,2 The record shows a durable and heavily used system, with turnout that swings widely by issue, a decades-long exclusion of women, outcomes that some see as limiting minorities, and a signature-collection integrity problem that federal authorities have been investigating since 2022.⁠^3,4,13,14

The comparative overview at /research/cases/ introduces Switzerland in a few paragraphs. This page goes deeper into the legal mechanics, the double majority, the 2014 immigration vote and its implementation, the cantonal open-air assemblies, the forged-signature investigations and the research record.

What happened

Swiss direct democracy at the federal level is a history of extensions and contested results. The timeline below lists events verified for this page.

• 1959. A federal vote on women’s suffrage was “rejected by 66.9% of voters and an overwhelming majority of the Swiss cantons,” according to the Law Library of Congress.⁠¹⁷
• November 15, 1970. A federal decree on the federal finance system received 55.38% yes but only 9 of 22 cantonal votes, so it was rejected.⁠⁵ It is a clear example of the double majority in action: a popular majority was not enough.
• February 7, 1971. The all-male federal electorate approved women’s suffrage in federal matters with 65.73% yes and 15.5 of 22 cantonal votes, on turnout of 57.73%.⁠⁴
• 1971 to 1972, Glarus. The last “Men’s Landsgemeinde” in Glarus, held in 1971, granted women full political rights in cantonal matters, and women have voted at the assembly since 1972.⁠¹⁵
• 1990 to 1991, Appenzell Innerrhoden. The Federal Supreme Court held in 1990 that excluding women from cantonal votes violated the equality clause of the Swiss Constitution, and women in the canton gained the cantonal vote in 1991.⁠¹⁷
• November 29, 2009. The initiative “Against the construction of minarets” passed with 57.50% yes and 19.5 of 23 cantonal votes, on turnout of 53.76%. Basel-Stadt, Geneva, Neuchâtel and Vaud rejected it.⁠⁶
• February 9, 2014. The initiative “Against mass immigration” passed with 50.33% yes and 14.5 to 8.5 cantonal votes, on turnout of 56.57%.⁠⁷
• June 20, 2014. A federal implementation concept stated that treaties conflicting with the new Article 121a, including the free movement agreement with the EU and EFTA, had to be renegotiated by February 9, 2017.⁠¹⁰
• December 16, 2016. Parliament adopted legal changes to implement Article 121a, including measures for job seekers in Article 21a of the Foreign Nationals Act.⁠¹¹
• July 1, 2018. A job vacancy reporting obligation took effect, implemented at ordinance level by a Federal Council decision of December 8, 2017.⁠¹²
• September 27, 2020. The “Limitation Initiative,” which would have ended free movement with the EU, was rejected with 38.29% yes and 3.5 of 23 cantonal votes, on turnout of 59.49%.⁠⁸
• October 2022 to January 2026. Federal authorities received criminal complaints about suspected forged signatures on initiative lists, beginning in October 2022. By January 2026 about 30,000 suspect signatures across about twenty federal initiatives had been reported, and the Federal Prosecutor’s Office and fedpol searched collection firms.⁠¹⁴

How it worked

Swiss federal direct democracy rests on a few articles of the Federal Constitution. Quotations below come from the English translation published by the Federal Chancellery, which states that it has no legal force.⁠²

Who can vote

Article 136 gives federal political rights to Swiss citizens aged eighteen or older, with an exception for people who lack legal capacity due to mental illness or incapacity.⁠² Residents without Swiss citizenship do not vote in federal matters under this article.

Popular initiatives

Under Articles 138 and 139, “any 100,000 persons eligible to vote may within 18 months of the official publication of their initiative” propose a complete or partial revision of the Constitution.⁠² An initiative may be a general proposal or a specific draft.⁠² The Federal Assembly must declare an initiative invalid, in whole or in part, if it fails the requirements of consistency of form or subject matter, or if it infringes mandatory provisions of international law.⁠² The check is limited to those grounds.

Counter-proposals

“The Federal Assembly shall recommend whether the initiative should be adopted or rejected. It may submit a counter-proposal to the initiative”.⁠² Under Article 139b, voters may approve both the initiative and the counter-proposal, and then answer a third, deciding question about which they prefer. If one proposal wins the popular vote on that question and the other wins the cantons, the one with the higher sum of percentages takes effect.⁠² The federal information service notes that authorities sometimes offer a direct counter-proposal hoping that “a majority of the people and the cantons support that instead”.⁠¹

Referendums

Article 140 states that certain items “must be put to the vote of the People and the Cantons,” including constitutional amendments, accession to organizations for collective security or to supranational communities, and emergency federal acts without a constitutional basis that last more than one year.⁠² An initiative in the form of a specific draft is likewise “submitted to the vote of the People and the Cantons”.⁠² Article 141 sets the optional referendum: a vote is held “if within 100 days of the official publication of the enactment any 50,000 persons eligible to vote” or “any eight Cantons request it”.⁠² Federal acts and certain treaties fall under this rule.⁠²

The double majority

Article 142 sets the decision rules. A proposal submitted to the people alone passes with “a majority of those who vote.” A proposal submitted to the people and the cantons needs a majority of those who vote and “a majority of the Cantons,” where “the result of a popular vote in a Canton determines the vote of the Canton”.⁠² Six cantons, including Obwalden, Nidwalden, Basel-Stadt, Basel-Landschaft and the two Appenzells, each have half a cantonal vote.⁠² The cantonal majority gives small cantons weight beyond their population, which is why the 1970 finance decree failed with 55.38% popular support.⁠⁵

Cantonal assemblies: the Landsgemeinde

The Landsgemeinde is an open-air assembly of eligible voters that still serves as the highest authority in Glarus and Appenzell Innerrhoden. A Glarus government publication describes it as follows:⁠¹⁵

• The regular assembly meets in the town of Glarus on the first Sunday in May. An extraordinary assembly is held if the assembly decides, if at least 2,000 electors request one, or if parliament calls one.
• The agenda is limited to motions published by parliament, but every voter may move to support, amend, reject, postpone or withdraw an item. The electorate “can eliminate provisions and amend or augment motions,” which the document calls unparalleled at cantonal level.
• “The Landammann determines the majority by estimation. In cases of doubt he can call on the help of remaining members of the government. His decision is final.”
• The same document lists disadvantages: voting is open rather than secret, not everyone who could vote by post can attend, weather affects participation, and estimating close majorities is a shortcoming.

National Geographic reported in 2025 that Glarus voters chose in 2007 to “lower the voting age from 18 to 16,” making Glarus the only Swiss canton where 16-year-olds can vote.⁠¹⁶

Measured results

Swiss turnout varies more by issue than by era. The Federal Statistical Office (BFS) states that participation in federal votes declined over the twentieth century with strong swings, that since 1990 the extreme values were 28% and 79%, and that average participation rose slightly from 42% in the 1980s and 1990s to 46%.⁠³ The Fribourg cantonal statistics office reports that 72 federal proposals went to a vote between 2020 and 2023, with nationwide average turnout of 51.7%.⁠²³ The federal information service describes recent decades as averaging “a little over 40%”.⁠¹ These figures use the registered Swiss electorate as the denominator, which excludes foreign residents.

Low average turnout does not mean most citizens never vote. A peer-reviewed study by Sciarini, Cappelletti, Goldberg and Lanz in the Swiss Political Science Review examined selective participation.⁠²⁰ Swissinfo, reporting on that research in 2016, said that in Geneva data covering the last 20 votes, “more than 90% voted at least once,” and “it was only around 10% who did not vote a single time”.⁠²¹ Selective voting means each result reflects whoever was mobilized by that issue.

Specific outcomes, all from Swissvotes, the University of Bern database:

Sources:.⁠^4,5,6,7,8

Implementation of the 2014 immigration vote

The 2014 result is the clearest test of what happens when a binding vote collides with treaty obligations. The new Article 121a requires that immigration be limited by annual caps and quotas and that Swiss residents be given priority, and its transitional provision required implementing law by February 9, 2017.⁠¹⁰ The federal implementation concept said the free movement agreement with the EU had to be renegotiated, noted that the Croatia protocol to that agreement could no longer be signed in its existing form, and recorded “a temporary blocking of all cooperation dossiers by the EU” before the Federal Council took measures on April 30, 2014.⁠¹⁰ The same document observed that Article 121a leaves the size of the reduction open, and that the initiators “never” specified it.⁠¹⁰

Parliament’s December 16, 2016 law chose a labor-market measure rather than caps on EU citizens.⁠¹¹ From July 1, 2018, employers in occupations with high unemployment had to report vacancies to public employment services, a measure the State Secretariat for Migration describes as serving “to combat unemployment”.⁠¹² A SECO report estimated that a 5% unemployment threshold would have covered 88 of 383 occupation codes in 2016.⁠¹¹

Failures and criticisms

Swiss direct democracy draws criticism from several directions, and each has a serious version.

• Exclusion by majority vote. Women were excluded from federal votes until 1971, and in one canton until a court ruled in 1990.⁠^4,17 Critics argue that a system in which the included decide whether to include others will be slow to correct itself. Defenders note that the 1971 extension was itself voted by the people.⁠⁴
• Minority rights. Hainmueller and Hangartner, in a 2019 American Journal of Political Science article using data from about 1,400 municipalities from 1991 to 2009, found that “naturalization rates surged by about 60% once politicians rather than citizens began deciding on naturalization applications”.⁠¹⁸ They argue that referendum voters “face no cost of arbitrarily rejecting qualified applicants,” while councils must justify rejections and face judicial review, and that the effect was stronger for more marginalized groups.⁠^18,19 The minaret result is often cited in the same debate.⁠⁶ Supporters of direct democracy respond that the initiative-invalidity rule protects mandatory international law.⁠² and that voters have also rejected restrictive proposals, as in 2020⁠⁸
• Implementation as override. Supporters of the 2014 initiative argued that Parliament did not implement it. The Swissvotes summary of the 2020 vote records the initiators’ view that the 2014 vote and “the very restrained implementation by Parliament” brought no improvement.⁠⁹ Opponents argue that Parliament had to weigh the 2014 article against the free movement agreement, which a guillotine clause ties to the rest of the Bilaterals I package, and the Federal Council called that agreement crucial for the Swiss economy.⁠⁹ Both readings are about who has the last word after a binding vote.
• Cantonal overweighting. The double majority means a popular majority can lose, as in 1970.⁠⁵ Federalists see this as protection for smaller regions; critics see unequal vote weight.
• Signature integrity. The Federal Chancellery filed its fifth criminal complaint in December 2025, covering about 9,500 signatures for five initiatives, most already declared invalid by municipalities.⁠¹³ Since early 2024 it has tightened counting controls, run a nationwide reporting system since mid-November 2024 and drafted a code of conduct.⁠¹³ The January 2026 federal statement on searches does not say that any initiative qualified only because of forged signatures.⁠¹⁴
• Open voting in assemblies. The Glarus government itself lists open voting and estimated majorities as weaknesses of the Landsgemeinde.⁠¹⁵ The same document argues that the assembly’s power to amend motions is unparalleled at cantonal level.⁠¹⁵
• Selective and unequal participation. Average turnout under half;⁠³ means each result reflects an issue-specific electorate. Research on selective voting shows that most citizens vote sometimes rather than never⁠^20,21 critics still see results driven by whoever is most motivated by a given issue.

What transferred and what did not

Swiss practice has shaped debates elsewhere, but the evidence verified for this page about direct transplants is limited. What can be documented is what Switzerland itself has declined to extend. The Federal Chancellery states that since 2004, “15 cantons have offered internet voting to a limited part of their electorate in over 300 trials,” that e-voting is permitted only if strict federal requirements are met, and that basic licenses cover only “a limited section of the electorate”.⁠²² Switzerland, in other words, kept its high-frequency voting on paper and by post, and treated remote electronic voting as a bounded trial. Estonia took the opposite path; see /cases/estonia-internet-voting/.

The Landsgemeinde has not spread: the Glarus government notes that only Appenzell Innerrhoden retains one, and that an assembly with representative participation is “only possible in a canton that has a small population”.⁠¹⁵ Citizens’ assemblies elsewhere draw on a different tradition, random selection; see /cases/ireland-citizens-assemblies/.

Lessons for My Vote Forward

The following points are interpretation and design proposals, not findings about My Vote Forward, which has not been tested.

• Separate agenda power from decision power. Switzerland’s thresholds let a minority (100,000 signatures) put a question on the agenda, while a majority decides. A framework that ranks public priorities is closer to agenda setting than to deciding; see priorities and proposals.
• State the second majority in advance. The double majority shows that “who won” depends on aggregation rules. Any published result should state its rule and show alternatives where they would differ; see tradeoffs.
• Decide what a majority cannot decide. The suffrage history and the naturalization study,⁠¹⁸ argue for limits stated before any vote, enforced by someone other than the majority; see representation.
• Plan for implementation gaps. The 2014 vote shows that even binding results are implemented by others. Results should be published alongside what was done with them; see participation over time.
• Treat collection integrity as a security problem. The forged-signature cases show that the step before a vote can be attacked; see identity and registration and audits and verification.
• Open voting has costs. The Landsgemeinde trades secrecy for public deliberation. My Vote Forward separates public opinion from private voting for this reason; see private voting and intimidation.

Unresolved questions

• How the Federal Act on Political Rights governs signature verification by municipalities in detail was not verified for this page, because the official text could not be retrieved.
• Did any federal initiative or referendum qualify only because of forged signatures? The federal statements verified here do not say so⁠^13,14 and the investigations were ongoing as of January 2026.
• Would the naturalization findings,⁠¹⁸ generalize to national votes on minority issues, or are they specific to municipal decisions on named applicants?
• How much of the turnout rise in the 2020 to 2023 period⁠²³ reflects particular high-salience votes rather than a durable change?
• Was the 2016 implementation law a faithful reading of Article 121a or a substitute for it? Federal documents record the open target⁠¹⁰ and the initiators’ view is recorded,⁠⁹ but no court ruling on that question was verified for this page.
• What share of initiatives receive a direct counter-proposal, and how often do counter-proposals win? Federal Chancellery statistics could not be retrieved for this page.