Estimation of comparative disease burden in epidemiological surveys is complicated by the fact that high comorbidities exist among many chronic conditions. The easiest way to take comorbidity into consideration is to distinguish between pure and comorbid conditions and to evaluate the incremental effects of comorbid conditions in prediction equations. This approach is illustrated here in an analysis of the effects of pure and comorbid major depression (MD) and generalized anxiety disorder (GAD) on a number of different measures of role impairment in the US National Comorbidity Survey (NCS) and the Mental Health Supplement to the Ontario (Canada) Health Survey (the Supplement). Pure MD and pure GAD were found to have roughly equal independent associations with role impairments. The incremental effects of having comorbid MD and GAD were found to vary depending on the outcome under investigation. The paper closes with a discussion of the methodological complexities associated with generalizing to comorbidities that involve rare conditions or more than two disorders. Copyright © 2002 Whurr Publishers Ltd.