This paper will present findings from experiments on mode of data collection and non-response that were embedded in an epidemiological! study on the age of onset of alcohol use and related behaviors. The experiments were an attempt to measure the impact of telephone versus web data collection modes and simultaneously measure the impact of non-response bias through a non-response follow-up survey of study non-participants. In addition, the study involves a comparison of findings to the National Epidemiologic Study on Alcohol and Related Conditions, a national benchmarking survey conducted by the Census Bureau by personal interviewing. The Age of Alcohol Onset Survey was a mixed-mode survey (web panel and telephone) of adults age 18-39. The sample design involves the administration of the some in-depth instrument to web panel members by Internet (n=3,409), web panel members by telephone (the control group n = 330), and non-responders to the web panel by telephone (n = 603). We examine three methodological issues: (i) the mode effects of data collection (internet vs. telephone); (ii) self-reported behavioral and attitudinal differences between responders and non-responders as measured by a non-response follow-up survey of persons who did not join the web panel or else dropped off the web panel; (iii) comparison between the web, telephone, and face-to-face interviews (the latter represented by NESARC). There were no significant differences in most of the major alcohol-related behaviors between the online panel and the NESARC results. The data collection mode effect was negligible for alcohol behaviors. The mode effect was more pronounced for attitudinal questions. Results of differences between the control group and the telephone and Internet modes will be reported. The study was conducted by Knowledge Networks on behalf of the Vauth Alcohol Prevention Center at the Boston University School of Public Health, and funded by the National Institute on Alcohol Abuse and Alcoholism.