Data on the prevalence of chronic morbidity are valuable for identifying health care needs and for designing services to meet these needs. Morbidity data are frequently collected through surveys based upon self-reports of disease. However, there is evidence in the literature that people tend to underreport the presence of chronic disease. Our study identified factors related to the underreporting of chronic disease. We interviewed a proportional stratified sample of 521 sick-fund members from two typical urban primary care clinics in Israel. The respondents' self-reports were compared to the diagnoses recorded in medical sources of information. Our findings show that more than half of the respondents with chronic diseases failed to report at least one disease. By using multivariate analysis, we were able to untangle the interrelated effects of age and multiple disease. The findings indicate that "number of diseases"--a variable not taken into account in previous studies--had the greatest effect on underreporting. When controlling for number of diseases, age had an independent effect, although a much smaller one. We also found that the rate of underreporting varies by type of disease. We conclude that chronic morbidity estimates based upon self-reports may lead to the deficient planning of health care services. If possible, alternative sources should be used for collecting morbidity data, especially from patients with multiple diseases and from the elderly. Further research is needed in order to improve estimates of chronic morbidity.