In a prospective longitudinal study of adolescents' drug use, the trustfulness of the self‐reports about drug use was investigated by means of a test for logical consistency. A sample of 1936 high school students completed a written survey in the fall of 1987 and again in spring 1988, seven to eight months later. The findings indicate that the information obtained is logically consistent at one particular point in time. Analyses show, however, that the consistency drops somewhat over time. The legal drug use responses show the highest consistency, the answers regarding hard drugs like amphetamine and heroin show the least favourable consistency, with cannabis and inhalants somewhere in the middle. There seem to be two main reasons for the inconsistent answers: some respondents wilfully underreport their drug consumption. This seems partly to be a function of the level of illegality of the drug. However, poor memory and episodic uses of such drugs seem also to be important. Generally this study agrees with other studies and indicates an overall relatively high level of longitudinal consistency regarding drug use responses. This kind of survey must therefore be regarded as a relatively reliable instrument in collecting information regarding drug us