Evidence-based policymaking (EBP) has become a global public management movement to improve constituents’ lives through government decision making. However, how civil servants’ decisions are influenced by such evidence remains unanswered. In this study, we answer two related research questions: 1) How do different elements of evidence impact civil servants’ program preferences? 2) How does the rating of evidence influence their program preferences? Collaborating with major governmental and nonprofit agencies that promote the use of EBPs, we invited civil servants from three U.S. state governments to a paired conjoint survey experiment. Our analysis shows that: Civil servants prefer programs with evidence that is: 1) from their own states; 2) more recent; 3) shows positive effect on people from different demographic groups; and is 4) created by independent government teams and university research teams. We find causal evidence that an “evidence-based” rating drives civil servants’ preferences toward evidence with higher internal validity.