Ballot order effects, whereby candidates or parties listed at the top of the ballot gain an electoral advantage, are well-documented across various majoritarian countries and electoral arenas. Yet, these effects remain relatively underexplored in proportional representation systems, particularly when party order is determined randomly. This study addresses this gap by utilizing a natural experiment in Portuguese legislative elections from 1979 to 2022, where the randomization of parties’ order on the ballot allows for the analysis of its causal effect on vote share. The dataset covers all 20 constituencies in Portugal, encompassing over four decades of electoral data. For each election and district, parties were assigned random ballot positions, offering the opportunity to isolate the effect of ballot order from other confounding factors. The findings, based on fixed-effects regression models, reveal that parties appearing at the top of the ballot achieve marginally better vote shares compared to those listed further down. This suggests that ballot order, even in proportional representation systems where party identity matters more than candidates’ visibility, can still shape voter behavior. These results highlight the need for further inquiry into how seemingly minor institutional factors, like ballot design, might influence democratic outcomes. This study not only contributes to the literature on electoral systems and voting behavior but also has potential implications for policymakers seeking to enhance electoral fairness. By showing the existence of causal ballot order effects in Portugal, it underscores the importance of randomization in mitigating structural biases.