Written by Tzu-chiao Su.
Image credit: 朱立倫/ Facebook.
On July 26, 2025, Taiwan witnessed an unprecedented event: a mass recall vote targeting sitting members of the Legislative Yuan. On that day, 24 district legislators from the Kuomintang (KMT)—the largest opposition party in the legislature—faced recall votes. This meant that two-thirds of the KMT’s 36 district legislators were at risk of being removed from office. The final result surprised many observers: not a single legislator was successfully recalled. Civic groups and the ruling party (the Democratic Progressive Party, DPP), who had hoped to use the recall process to restructure party seat distribution in the legislature and thereby change the political deadlock between a divided government and a “minority president–majority opposition” configuration, ultimately failed in their endeavour.
Under Taiwan’s current recall system, if 1% of eligible voters in a single-member legislative district initiate a petition and 10% of the electorate support it via signature, a formal recall vote is triggered. For the recall to pass, the number of “yes” votes must exceed the “no” votes, and the total “yes” votes must reach at least 25% of all eligible voters in that district. If these conditions are met, the legislator is removed from office and a by-election is held to fill the seat.
Within such a framework, many KMT legislators facing recall felt deeply anxious. Pre-election polling suggested that roughly 60% of voters in the affected districts opposed the recalls, while about 40% supported them—indicating more opponents than supporters. However, general expectations about voter behaviour cautioned that recall supporters would be more motivated to vote, while opponents might choose to stay home. In other words, despite stronger opposition to the recall in principle, asymmetric mobilisation could still lead to a successful outcome for recall proponents.
For example, in the 2017 recall vote against then-legislator Huang Kuo-chang and the 2022 recall vote targeting then-legislator Freddy Lim, although pre-election polls at the time indicated that the proportion of voters opposing the recall of these legislators was not lower than those supporting it, the final voting results showed that the number of votes in favour of recall exceeded those opposed in both cases. However, both efforts ultimately failed because the turnout fell short of the legally required threshold of 25% of eligible voters. A review of previous recall cases in Taiwan reveals a consistent pattern: since those opposing the recall are typically less motivated to vote, overall turnout in such elections tends to be low—usually ranging between 40% and 50%, and in some instances dropping below 30%, as was the case in the 2017 Huang Kuo-chang recall attempt.
Yet the result of the mass recall vote was unexpectedly clear: Voter turnout in this recall election was significantly higher than in previous recall attempts, with most districts recording participation rates between 55% and 60%. In all 24 districts, the number of “no” votes exceeded “yes” votes. Although in 7 districts the “yes” votes surpassed the 25% turnout threshold, they still fell short due to being outnumbered. Notably, coincidentally matching the number above, there were also seven districts in which the number of “no” votes surpassed the votes the incumbents had received during the previous legislative election. In 23 districts (with only one exception among the total districts), the total “yes” votes—generally representing support from the ruling party or the pan-Green camp—fell short of the vote count President Lai Ching-te had received in the 2024 presidential election, indicating that even many Green supporters did not back the recall efforts.
Several observations emerge from this electoral outcome:
First, the campaign’s nationalist narratives promoted by the recall groups and the ruling party—especially themes such as “resist China, protect Taiwan” and “oppose communism, defend Taiwan”—failed to resonate. Voters in these districts largely rejected the framing of KMT legislators as “collaborators of the Chinese Communist Party.” Such grand and morally charged narratives, which might be effective in presidential campaigns, do not align with voter expectations at the legislative level. Constituents tend to evaluate legislators based on district service and parliamentary performance, not on abstract notions of national defence or ideological loyalty. The fact that “yes” votes in nearly every district fell below President Lai’s vote count from the previous year suggests that even Green-leaning voters were unconvinced.
Second, the results reveal a strong “spiral of silence” effect in public opinion. Throughout the recall campaign, public discourse—particularly on social media and in street rallies—was dominated by pro-recall voices. Online slogans such as “all sectors support the recall” created the illusion of mainstream consensus. Consequently, many opponents of the recall opted to remain silent. However, the final vote count made clear that the true majority had simply been the silent one.
Third, in several districts, the number of “no” votes exceeded the vote counts these legislators had received in the previous election. This suggests that supporters of both the KMT and the Taiwan People’s Party (TPP) may have jointly opposed the recalls. Over the past year, the much-discussed “blue-white alliance” between these parties has extended beyond elite-level cooperation and begun to manifest among grassroots voters.
Fourth, the results vividly illustrate what political science terms Fenno’s Paradox: voters generally dislike the legislature as a whole but maintain favourable views of their own district representatives. Even if, at a national level, some voters were sympathetic to the ruling party’s accusations of “legislative chaos” or “pro-China pandering” over the past year, they were ultimately reluctant to oust their local legislators—especially in districts where non-Green voters constitute the majority.
Globally, very few democratic nations allow for the recall of national legislators. This is grounded in prevailing theories of representative democracy, which posit that elected legislators not only represent their local constituencies but also serve the national interest. Once elected, legislators are entrusted with the autonomy to exercise independent judgment in pursuit of the public good and should not be excessively constrained by immediate district preferences. Overuse of recall mechanisms can deter legislators from making difficult decisions, pressuring them to pander to parochial interests and short-term populism. Moreover, recall campaigns often involve negative or even hostile mobilisation strategies, increasing the risk of political polarisation. For these reasons, most democratic systems do not implement legislative recall procedures.
Taiwan’s mass recall movement briefly opened a political Pandora’s box. Fortunately, the results of the vote managed to close it again. This year’s wave of recall campaigns and their outcomes showcased not only Taiwan’s vibrant democratic energy but also its strong democratic resilience. Yet the ultimate failure of the recall movement may also have a paradoxical consequence: it could deprive Taiwan of the opportunity to critically reassess its recall mechanism, including its arguably low threshold for success. This may be the most ironic and enduring implication of the 2025 mass recall episode.
Tzu-chiao Su is a Professor at the Department of Political Science at Soochow University, Taiwan. His research focuses on constitutional operations across different countries and the development of democracy worldwide.
This article was published as part of a special issue on ‘Recall elections: Practice or problem for Taiwan’s democracy?‘.
Taiwan Insight 