Furthest from power, most trusted
Trust in Serbia sorts along a single fault line - distance from party politics. The two institutions citizens trust most are the ones least entangled in day-to-day government: state universities (71%) and the Serbian Orthodox Church (63%). The police, civil-society organisations and the judiciary sit in the low-to-mid forties. And at the very bottom are the elected political institutions - the President and the Government at around a third (35% and 34%), the National Assembly at 31%, and political parties at just 19%. For a democracy this is an inversion: the bodies meant to embody the popular will are the least believed in, precisely because they are seen as captured by one side.
The presidency's steep fall
That low base is the result of a steep, recent slide, sharpest at the top. Trust in the institution of the President has nearly halved in three years - 60% in February 2023, 40% in April 2025, 35% now - the largest fall of any institution measured. The Government has dropped from 49% to 34% and the Assembly from 36% to 31% over the same period, while the church has edged up (59% to 63%). The erosion, in other words, is concentrated in the elected core of the state; the institutions that answer to voters are losing them fastest.
The state and the party as one
The averages hide the sharpest split: trust is almost entirely a function of which side you are on. Among ruling-party (SNS) voters, trust in every institution is high - 98% trust the President, 92% the Government, 88% the Assembly, even 94% the church and 90% the police. Among citizens who do not vote for the governing parties, trust in the political institutions all but vanishes: essentially none of the student-list voters trust the President (0%), the Government (1%), the Assembly (1%) or the parties (1%). Trust in the state has collapsed into a proxy for allegiance to the ruling party - wholly trustworthy to those who back it, and barely credible to everyone else.
That fusion is not only a matter of perception. After the December 2023 elections the OSCE/ODIHR found that the misuse of state resources had "blurred the line between state and the party," a finding its follow-up rated not implemented. It repeated the call after the 2024 local elections as a priority recommendation (1 (2024)): "measures should be taken to ensure the separation of the state and party and the impartiality of the public administration." What the observers describe from the institutional side, the survey shows from the citizens'.
That fusion is not only a matter of perception. After the December 2023 elections the OSCE/ODIHR found that the misuse of state resources had "blurred the line between state and the party," a finding its follow-up rated not implemented. It repeated the call after the 2024 local elections as a priority recommendation (1 (2024)): "measures should be taken to ensure the separation of the state and party and the impartiality of the public administration." What the observers describe from the institutional side, the survey shows from the citizens'.
A president known, a prime minister unknown
The personalisation of power shows in two more readings. Only about a third of citizens (36%) say they trust President Aleksandar Vučić, while 61% do not - distrust that runs highest among the young and reaches even citizens with no clear political camp.
And yet, asked simply to name the current prime minister, only 53% get it right (Đuro Macut); 41% do not know and 6% name someone else. Who knows and who does not follows engagement and education far more than party: recognition reaches 67% among the university-educated (against 36% of those with only primary schooling) and is roughly equal among committed voters on both sides - 61% of ruling-party voters and 62% of student-list voters - but falls to 39% among the undecided and just 21% among those who will not vote.
Media habits matter too, with audiences of N1 and Nova S (69%) and online portals (63%) more likely to know than viewers of the public broadcaster or the commercial channels (both around 54%). The citizens least able to name the head of government are the ones most detached from politics, not one camp or the other - and a public that can identify the president but not the prime minister is telling you where it believes authority actually sits.
And yet, asked simply to name the current prime minister, only 53% get it right (Đuro Macut); 41% do not know and 6% name someone else. Who knows and who does not follows engagement and education far more than party: recognition reaches 67% among the university-educated (against 36% of those with only primary schooling) and is roughly equal among committed voters on both sides - 61% of ruling-party voters and 62% of student-list voters - but falls to 39% among the undecided and just 21% among those who will not vote.
Media habits matter too, with audiences of N1 and Nova S (69%) and online portals (63%) more likely to know than viewers of the public broadcaster or the commercial channels (both around 54%). The citizens least able to name the head of government are the ones most detached from politics, not one camp or the other - and a public that can identify the president but not the prime minister is telling you where it believes authority actually sits.
Wary of others, not of each other
Two further readings round out the picture, and they measure different things. The first is general trust in other people: here Serbia scores low, with 56% saying you can never be too careful in dealing with others and just 21% that most people can be trusted - the kind of low-trust setting that tends to make collective action harder.
The second is tolerance of political difference, and it is higher than the polarisation might suggest: 73% of citizens say they would not mind socialising with people whose political views differ sharply from their own. So although trust in institutions has divided sharply along party lines, that animosity has not, for the most part, turned citizens against one another in daily life.
The second is tolerance of political difference, and it is higher than the polarisation might suggest: 73% of citizens say they would not mind socialising with people whose political views differ sharply from their own. So although trust in institutions has divided sharply along party lines, that animosity has not, for the most part, turned citizens against one another in daily life.
The through-line
Taken together, the trust findings have a consistent shape. Confidence has drained from the political institutions - and, for citizens outside the governing bloc, from the state itself - while trust in non-partisan institutions and everyday tolerance between citizens have held up better. The structural feature running through all of it is the blurred line between the state and the ruling party: for a large part of the public, the two have effectively become one.
Source: CRTA public opinion survey, conducted with the Democracy Action Lab (Stanford CDDRL); fieldwork by Pulse Research. Face-to-face (CAPI), n=2324, representative of the adult population of Serbia (excl. Kosovo), 10-24 June 2026, margin of error ±2.08%. The trust-over-time comparison draws on earlier CRTA surveys (February 2023, April 2025) using the same method. The social-trust and socialising findings are from the same June 2026 survey but are not included in the summary deck.
