OBJECTIVE: To determine whether self reported overall honesty in questionnaire responses differed by sociodemographic characteristics and/or extent of involvement in violent, delinquent, and high risk activities.DESIGN: As part of an urban district middle (MS) and high (HS) school based sexuality prevention program, all health class students in 9 participating schools were administered a 186 item pretest questionnaire. Fewer than 5% of eligible students declined participation, resulting in 2241 MS and 1609 HS subjects.RESULTS: The mean age for the total sample was 14.8 years (S.D. 2.5), with a range of 12 - 19 years. There were similar numbers of males (51.6%) and females (48.4%). The sample was ethnically diverse, with 50% African-American, 21% Hispanic, 16% white, and 13% other/mixed racial; 85% of subjects were from families at or below the federal poverty threshold. For the total sample, 40% of subjects reported ever carrying a weapon, 67% reported having been in a physical fight, and 12% having been arrested. Using a 7 point visual analogue scale from not at all (1) to completely (7) honest in overall questionnaire responses, 85% of MS and 90% of HS subjects reported answering very honest (score 6 or 7), with 3.4% of MS and 2.0% of HS subjects reporting as dishonest (score 1-3). For sexual behaviors, 67% of subjects in MS and 81% in HS reported answering honestly, but significantly more MS males than females overreported their behaviors (23% vs. 12.6%, p<.001). Similar numbers of MS males and females reported underreporting their sexual behaviors (16% vs. 15.2%, p=NS). Males and female high school students did not differ in their honesty about sexual behaviors, with 10.8% of males and 8.1% females overstating, and 10.1% males and 9.1% females understating their behaviors. A Life Risk Index was created using 17 high risk behaviors, including substance use, risky recreational vehicle use, delinquent activities, and violent behaviors; a separate Violence Risk Index was created using four behaviors (weapon-carrying, physically fighting, and being arrested or convicted). Reported overall honesty was assessed for those scoring in the highest 25%ile (greater involvement in behaviors) vs. the remainder; overall honesty was not significantly associated with extent of involvement in either general risk behaviors or specific violent behaviors. Using logistic regression, and controlling for socioeconomic status, ethnicity, and involvement in high risk and violent behaviors, we found that females were twice as likely (p<.001) and high school students 1.7 times (p<.001) more likely to report being honest (score 6 or 7).CONCLUSIONS: The findings indicate high levels of reported overall honesty among middle and high school students, with less honesty reported for sexual behaviors. While those reporting less overall honesty were more likely to be younger and male, there was no association with extent of involvement in violent or high risk behaviors. The data lend support to relying upon adolescents' self reported survey responses.