Jury decisions are among the most consequential social decisions in which bias plays a notable role. While courts take a number of measures to reduce the influence of bias on decisions about case strength or deserved punishment based on evidence introduced during a trial, jurors may still incorporate personal biases based on knowledge, experience, emotion, and beliefs independent of evidence. One common form of this bias, crime-type bias, is the extent to which the perceived strength of a case depends on the severity of the crime. A number of explanations from psychology and law point to the role of moral judgment, social cognition, and affect as core processes of bias. However, behavioral evidence alone makes these explanations difficult to distinguish. To overcome this challenge, we used fMRI to record brain activation patterns of mock jurors as they read a series of criminal scenarios and rated the strength of the cases and deserved punishment. Compared to patterns of brain activation derived from large neuroimaging databases, mock jurors’ neural activation patterns related to crime-type bias were most similar to patterns associated with social cognition (such as those associated with mentalizing and racial bias) but not affect or moral judgment. Further, results indicated that crime-type bias could be explained by variability in victim harm. Our results support a central role for social cognition in juror decision making and suggest that crime-type bias may arise from similar mechanisms that precipitate other biases like stereotypes about culture or race.