Study data were obtained from a comparative study of service delivery by 37 police agencies in the St. Louis, Mo.; Rochester, N.Y.; and St. Petersburg, Fla., metropolitan areas. The data are analyzed on three levels: (1) the individual officer, (2) individual police-citizen encounters, and (3) the neighborhood, aggregating characteristics of officers, citizens, and encounters. This permits examination of individual differences in attitudes among more and less educated sergeants and among command rank officers. The analysis also considers the effects of the educational characteristics of sergeants and commanders on (1) the attitudes of individual patrol officers, (2) the relationship between individual patrol officer education and attitudes, (3) individual patrol officer performance, (4) aggregate attitudes of patrol officers serving a study neighborhood, and (5) aggregate patrol officer performance. Individual encounters between patrol officers and citizens were observed and recorded by trained, nonpolice observers in the areas of citizen anger, police labeling, force, verbal abuse, and helping, to measure police performance. Other data collection sources were inperson interviews with more than 1,800 officers in 37 departments and a telephone survey of about 12,000 citizens in 60 neighborhoods under the jurisdictions of 24 departments. Findings show the now familiar pattern of no relationship between commander education and patrol officer perfomance; however, considerable evidence was found of relationships between commanders' education levels and their own attitudes and the attitudes of officers under them. These findings are dampened, however, by the failure of police studies to document a link between attitudes and performance. Tabular data, notes, and a bibliography of 19 references are provided.