Respondents were asked a battery of questions about their experience with each substance, recency of use, frequency, level or dosage, and indicators of abusive use. Three types of longitudinal inconsistencies were examined: retractions, modifications, and new admissions. Less than five percent of the students provided incomplete or inconsistent responses within questionnaires. Most discrepancies were attributable to unintentional errors caused by carelessness or misunderstanding rather than deliberate distortions. More than 40 percent of the students committed at least one longitudinal inconsistency across the 4 waves of data collection. However, over 95 percent of these errors were minor -- classifiable as either inconsistent reporting of infrequent experimental use or as inaccurate recall of whether limited prior use had occurred within the past year. Severe inconsistencies, which involved denying frequent use, were committed by less than two percent of the students. Laboratory assessments of saliva provided independent verification that the majority of students accurately reported recent cigarette use. It was concluded that the overwhelming majority of Project ALERT students accurately reported their drug use.