Due to limits on data availability in off census years, the study analyzed the determinants of police per capita strength in 1980, 1990, and 2000 for U.S. cities with populations greater than 100,000 in 1980. In contrast to the random-effects approach used in most panel analyses, the fixed-effects strategy used in the current study holds constant any unchanging case attributes by entering separate dummy variables for each city. This makes fixed-effects estimates unbiased when unmeasured time-invariant city characteristics associated with the explanatory variables influence the dependent variable. The study measured racial threat with the natural log of the percentage of African-Americans in a city. Ethnic threat was measured with the percentage of Hispanics. Residential segregation was assessed with the index of dissimilarity based on the odds that any Black will come into contact with any White. Period effects that may influence police department strength across all cities in a particular year were held constant by including two dummy variables coded 1 for 1990 and 2000. The findings show that the link between racial threat and the population-corrected number of police officers has recently become significantly stronger. Tests for interaction showed that segregated cities with larger African-American populations had smaller departments. The fact that racial segregation leads to reductions in police strength in the South may be due to officers being less likely to intervene in residentially isolated Black neighborhoods in the South. Overall, however, this study concludes that police strength in the United States increasingly reflects the size of minority racial and ethnic groups in the jurisdiction. 5 tables, 1 figure, and 53 references