In this analysis of the high school senior classes of 1976 through 1979, reports of frequency of drug use during the past month are roughly three times larger than would be estimated based on reports of use during the past year; this phenomenon appears fairly consistently for alcohol, marijuana, arid 10 other categories of illicitly used drugs. The underreporting of events that are more distant in time has been observed in a wide range of surveys, and these findings are general and stable enough to fit in very well with that explanation. The authors conclude that self-reports of frequency of drug use during the past year, and also during the lifetime, are in many cases systematically underreported, that percentages reporting any use during a given interval arc likely to be more accurate and, further, that analyses of trends are likely to be largely valid, since biases are likely to be fairly constant from year to year.