The data analyzed were from the "State Court Processing Statistics, 1990-2000: Felony Defendants in Large Urban Counties" (U.S. Department of Justice, Bureau of Justice Statistics, 2004). Logistic and linear fixed effects models were used for the following five response variables: denial of bail, nonfinancial release, bail amount, made bail, and pretrial incarceration. The analyses were disaggregated by both decision and crime type. The dataset contained a representative sample of State felony cases in large metropolitan counties in the years 1990, 1992, 1994, 1996, 1998, and 2000; it provided detailed information on prior record and offense severity, a comprehensive list of common offenses, several measures of demographic characteristics, and a nationally representative sample of adequate size. The study found that racial disparity was most notable in the decision to deny bail and for defendants charged with violent crimes. Ethnic disparity was most notable in the decision to grant nonfinancial release and for defendants charged with drug crimes. When there was disparity in the treatment of Black and Latino defendants with similar legal characteristics, Latinos always received the less beneficial decisions. These findings are consistent with the theoretical perspective proposed, which suggests that decisionmakers' stereotypes of charged felons based on race and ethnicity influence criminal processing when their specific content is made salient by either the concerns relevant to a particular processing decision or the crime type of a defendant's primary charge. 5 tables and 72 references