The paper of Anderson et al. (1993), based on cross-sectional data, showed that minor psychiatric disorder in a population is linearly related to the mean number of psychiatric symptoms in the population. The present investigation asks whether the same relationship holds longitudinally as well as cross-sectionally. Data from a 7-year follow-up of a general population sample demonstrate, for the first time, that a relationship exists between changes in prevalence of psychiatric disorder and changes in the mean number of psychiatric symptoms in a given population. Moreover, the relationship is linear; a one-point decrease in mean scores on the GHQ-30 is associated with a 6% decrease in prevalence of disorder.