Background. Depression is highly co-morbid with both psychiatric and chronic somatic disease. These types of co-morbidity have been shown to exert opposite effects on underdiagnosis of depression by general practitioners (GPs). However, past research has not addressed their combined effect on underdiagnosis of depression. Method. Co-morbidity data on 19l depressed primary-care patients selected by a two-stage sampling procedure were analysed. Diagnoses of major depression and/or dysthymia in the last 12 months were assessed using a standardized psychiatric interview (CIDI) and compared with depression diagnoses registered by GPs in patient contacts during the same period. Presence of psychiatric and chronic somatic co-morbidity was determined using the CIDI and contact registration, respectively. Results. Regression analysis showed a significant interaction effect between psychiatric and chronic somatic co-morbidity on GPs' diagnosis of depression, while taking into account the effects of sociodemographic variables, depression severity and number of GP contacts. Subsequent stratified analysis revealed that in patients without chronic somatic co-morbidity, a lower educational level, a less severe depression, and fewer GP contacts all significantly increased the likelihood of not being diagnosed as depressed. In contrast, in patients with chronic somatic co-morbidity, only having no psychiatric co-morbidity significantly decreased the likelihood of receiving a depression diagnosis. Conclusions. Our results indicate that the effects of psychiatric co-morbidity and other factors on underdiagnosis of depression by GPs differ between depressed patients with and without chronic somatic co-morbidity. Efforts to improve depression diagnosis by GPs seem to require different strategies for depressed patients with and without chronic somatic co-morbidity.