Predictors of non-response were investigated in a 15-year follow-up (1981-1996) of 3,481 individuals in a probability sample from the household population of East Baltimore. Demographics (age, sex, race) and psychopathology (depression, anxiety, substance abuse) were examined as predictors of attrition. The study found that individuals with higher levels of psychopathology were more likely to drop out of the study, suggesting that psychopathology may be a significant predictor of attrition in longitudinal studies.