A recent study of police behavior using data collected in Dade County, Florida, found that citizen demeanor is a spurious correlate of arrest in light of control for criminal conduct (Klinger, 1994). This finding calls into question the long‐standing belief that hostility directly increases the odds of arrest in police‐citizen encounters. Responding to this research, Lundman (1994) reanalyzed data used in several previous studies that had reported hostility effects. His reanalysis offered limited support for a demeanor‐arrest link. Because the measures of demeanor he used and the models he estimated were somewhat different from those Klinger had used and estimated, Lundman suggested that it would be valuable to revisit the Dade County data to see whether Klinger's null finding regarding hostility effects might be artifactual. This study reanalyzes the Dade County data. It indicates that one of four measures of demeanor is a significant net correlate of arrest under some circumstances. The implications of this finding are disc