The study sample ranged in age from 13-years old to 16-years old and was made up of new admissions to a southern institution for juvenile offenders. The sample consisted of 40 subjects who were randomly assigned to 1 of 4 conditions, including 3 levels of attention and 1 control group. All subjects receiving the attention conditions were seen on a once, twice, or three times per week basis for 30-minute sessions for the duration of the study, for a total of 4 weeks for each subject. Data were collected on the runaway behavior and institutional offenses of each subject during the study. Prior to assignment to groups all subjects were administered a series of personality and demographic measures. Data were subjected to a multivariate correlational analysis and a Fisher Exact Test. Comparisons were made between experimental groups as well as experimental groups versus the control group. Results indicated that attention did significantly reduce running behavior, although a hierarchy was not found that suggested the value of one attention level over others. With institutional offenses behavior a hierarchy was not found, although a significant difference was found when the two highest levels of attention were compared with the control condition and the one time per week attention sessions. In future research it would be desirable to include measures of the emission of positive behaviors in addition to the measures used in this study. In general, findings support the value of attention in diminishing negative responses. Tables, appendixes of study and related data and a bibliography of 15 references are included. (Author abstract modified)