In Japan approximately 20 percent of high schools are for boys only, 35 percent are for girls only, and the other 45 percent are coed. In examining the influence of these various school environments on delinquent behaviors, role theory predicts that, because of feminine-macho role polarization, sex segregation has a positive effect on girl's delinquency and negative effects for boy's delinquency. In contrast, the Differential Association Theory predicts negative effects on girl's delinquency and positive effects for boy's delinquency. The study's questionnaire included a translation into Japanese of much of Hirschi's Richmond Youth Study instrument for administration to high school students. Additional items tested routine activity as well as social control theory. For the dependent variable of delinquency, six crime items were used: vandalism, battery, small theft (less than $2 value), medium theft ($2-$50), large theft (more than $50), and joy riding with someone else's bicycle or motorcycle without permission. Eight items on status offenses were added as dependent variables to cover acts regarded as improper in Japan. Self-reported delinquency data for 1,122 girls were collected in Osaka in 1988. For adult criminal offenses, Role Theory was correct in predicting a positive impact for sex-segregated schools for girls; for status offenses, however, Differential Association Theory was right in the prediction of a negative effect for sex-segregated girls schools. 3 tables and 8 references