OBJECTIVE: To characterize the prevalence of weapon carrying in an epidemiologic sample of about 1500 urban schoolchildren between spring 1989 and spring 1993. DESIGN: A descriptive analysis based on 5 years of cross-sectional confidential survey data. SETTING: An urban public school system in the mid-Atlantic region of the United States. PARTICIPANTS: The study base included all children who were interviewed in each year from 1989 to 1993, disregarding the number of prior or subsequent interviews. The mean age of the participants in the sample was 9 years when the survey began. MAIN OUTCOME MEASURES: Carrying a nonlethal or lethal weapon (eg, a stick, knife, or gun) to defend or to threaten or hurt someone, assessed by self-report. Reports of interpersonal aggression also were obtained. RESULTS: Knife and gun carrying increased with age while stick carrying decreased. In 1989, carrying a lethal weapon (eg, knife or gun) was reported by 11.7% of boys and 3.3% of girls; in 1993, these proportions had increased to 22.2% and 15.3%, respectively. Gun carrying increased over time; by 1993, 9.9% of boys and 1.4% of girls reported that they had carried a gun in the previous year. Carrying a weapon to defend or protect oneself was more commonly reported than to hurt or threaten someone. Various types of interpersonal aggression were frequently reported among boys and girls in each year. Our findings support the hypothesis that early involvement with a weapon is associated with later more serious involvement with a weapon. CONCLUSIONS: Involvement in weapons-related behavior starts young, well before the middle school year. Effective intervention programs may require attention to the early determinants of involvement with weapons, sex differences, and perhaps also the changing profile of reasons given for carrying and using weapons in conjunction with other forms of aggressive behaviors.