Abstract Epidemiological data show that the co-occurrence of two or more supposedly separate child (and adult) psychiatric conditions far exceeds that expected by chance (clinic data cannot be used for this determination). The importance of comorbidity is shown and it is noted that it is not dealt with optimally in either DSM-III-R or ICD-9. Artifacts in the detection of comorbidity are considered in terms of referral and screening/surveillance biases. Apparent comorbidity may also arise from various nosological considerations; these include the use of categories where dimensions might be more appropriate, overlapping diagnostic criteria, artificial subdivision of syndromes, one disorder representing an early manifestation of the other, and one disorder being part of the other. Possible explanations of true comorbidity are discussed with respect to shared and overlapping risk factors, the comorbid pattern constituting a distinct meaningful syndrome, and one disorder creating an increased risk for the other. Some possible means of investigating each of these possibilities are noted.