This study used merged data from three national surveys to address questions about security measures in American high schools, suspension rates, and student misbehavior. First, the study identified the characteristics of schools that implemented the most extreme security measures and those with the highest levels of discipline. Second, the study used data on individual students to examine misbehavior and race and gender disparities in suspensions not attributable to misbehavior. The main findings were: (1) In-school suspensions serve a 'gateway' function with regard to out-of-school suspensions. They may provide a time and place to address behavior problems before they escalate or disproportionate out-of-school suspensions before they occur; (2) Out-of-school suspensions were more frequent among schools in higher-crime neighborhoods. Thus students suspended may be relegated to an environment not conducive to positive educational or social outcomes; (3) African-American students and Hispanic/Latino students were suspended at higher rates than were non-Hispanic whites, differences in most cases not attributable to different levels of misbehavior; (4) Overall, males were more likely to be suspended than were females, an effect above and beyond that explained by differences in behavior. There was little or no difference in the suspension rates of black males and females, however; (5) High degrees of school security were associated with increased suspension rates and increased black-white disparities in total suspensions. At the same time, most black students were enrolled in schools with high degrees of security; and (6) Black males were suspended at higher and higher rates as school size increased. All of these have implications for school policy and practice. These are discussed in the paper together with limitations of the investigation.