PoderData shows that 52% think that the work of ministers is “bad” or “terrible”, the highest rate since 2021; only 9% say it is “good” or “great”
Brazilians’ perception of the work of ministers of the Federal Supreme Court reached the worst level since 2021, when the PowerData asked the question for the first time. Today, 52% of voters say that the work carried out by the Court’s ministers is “bad” or “terrible”. In June 2021, it was 31% – a jump of 21 percentage points in the period. In the last survey, carried out in 2025, the rate was 44%.
The percentage of respondents who say that the magistrates’ performance is “good” or “excellent” fell from 14% to 9% in 3 months, the lowest in 5 years. The positive perception of ministers was never very significant. It reached its peak (31%) in December 2022, after the election of the president (PT), but has always been close to 20%.
The research was carried out by PowerDataa company in the Poder360 Jornalismo group, with its own resources. Data were collected through calls to cell phones and landlines. There were 2,500 interviews in 132 municipalities in the 27 units of the Federation. The margin of error is 2 percentage points. The confidence interval is 95%.
To reach 2,500 interviews that proportionally fill (as they appear in society) the groups by sex, age, income, education and geographic location, the PowerData makes tens of thousands of phone calls. Often, there are more than 100 thousand calls until interviewees are found that faithfully represent the population as a whole. Find out more about the methodology by reading.
APPROVE AND DISAPPROV LULA
Among those who declare their disapproval of the PT member, only 3% say that the STF ministers do a “good” or “excellent” job. Those who think the performance is “bad” or “terrible” in this group total 61%.
In the group that says they approve of Lula, 18% of voters have a positive perception of the judges. The negative evaluation reaches 37%.
WHY DOES IT MATTER
Because the Supreme Court is going through its worst moment.
The negative evaluation reached 52%. It is the worst level in the historical series. The positive rate dropped to 9%. This is not just about oscillation, but about the loss of legitimacy perceived at an accelerated pace.
The breakdown by government assessment is revealing. Among those who disapprove of Lula, the majority reject the STF: 61% consider the work to be bad or terrible and only 3% evaluate it positively. Among the president’s supporters, there is more tolerance, albeit limited: 37% have a negative assessment and 18% a positive one.
The Supreme Court has long stopped being seen as a non-partisan institution. The relationship between the STF and Congress is increasingly worse. If this situation persists post-election, especially in a scenario without Lula as President, it increases the likelihood of advancement of initiatives currently contained, such as the opening of impeachment proceedings against ministers.
The Senate has never impeached a member of the Court. But this unprecedented development has never been so close to an end.
PODERDATA
The content of PowerData can be read on social media, where infographics and news are shared. Follow the profiles of the research division of Poder360 no, no, no and no.
Read other reports from this round:
SEARCH AGGREGATOR
O Poder360 offers those of Drive o , the oldest and most complete on the internet in Brazil. It brings together thousands of voting intention surveys from all companies since the year 2000. In election years, only studies that are registered with the Electoral Court and have a known complete methodology are published. Do you have any research to share? Email the full text to Poder360: .
If you are a subscriber, you can access the Research Aggregator and search for the data you want for disputes from 2026 or all previous years. Read how to sign Drive to access the Search Aggregator and other products from Poder360.
METHODOLOGY
The search PowerData was carried out from March 21 to 23, 2026. 2,500 people aged 16 or over were interviewed in 132 municipalities in the 27 units of the Federation. Parametric weighting was applied to compensate for disproportionalities in the variables of sex, age, level of education, region and income. The margin of error is 2 percentage points, plus or minus.
The interviews were carried out by telephone (for landlines and cell phones), using the URA system (Audible Response Unit), in which the interviewee listens to recorded questions and responds using the device’s keyboard. The study’s confidence interval is 95%.
To make reading easier, the search results have been rounded. Because of this process, it is possible that the sum of some of the results is different from 100%. Differences between total frequencies and percentages in variable crossing tables may appear due to non-response occurrences. This study was carried out with the company’s own resources. PowerDataa research company that is part of the media group Poder360 Journalism.
