Investigations of how criminal justice actors contribute to variation in sentencing typically focus on the role played by the judge. We argue that sentencing should be viewed as a collaborative process involving actors other than the judge and that the role of the prosecutor is particularly salient. We also contend that the courtroom workgroup literature has suggested that sentences may vary depending on the particular judge and prosecutor to whom the case is assigned. By using a unique data set from three U.S. district courts (N = 2,686) that identifies both the judge and the prosecutor handling the case, we examine how the judge, the prosecutor, and the judge–prosecutor dyad contribute to variance in offender sentences. We do this by employing cross‐classified random‐effects models to estimate the variance components associated with judges, prosecutors, and judge–prosecutor interactions. The results indicate that disparity attributable to the prosecutor is larger than disparity from the judge. Moreover, the role that the judge plays is moderated by the prosecutor to whom the case is assigned, as the judge–prosecutor effect is consistently larger than other random effects across the models. We also find that results vary by judicial district.