This study evaluates the quality of data obtained from after-death interviews with significant others of deceased older persons regarding the prevalence of chronic diseases and symptoms in the terminal phase of life. These data are compared with reports from physicians and earlier self-reports from the deceased person. There were significant increases in nonresponse and nonavailability of significant others for decedents who had been divorced or had never been married, thus introducing some selection bias. At the level of the total sample, significant others seem to give accurate information about the prevalence of chronic diseases when compared with self-reports and reports from physicians. At the level of the individual sample member, after-death interviews with significant others provide valid information for the assessment of the prevalence of malignant neoplasms, diabetes mellitus, chronic obstructive pulmonary disease and cerebrovascular disease, but not for osteo- and rheumatoid arthritis and artherosclerotic disease. At the level of the total sample, the prevalence of symptoms assessed by significant others did not differ greatly from the assessment made by physicians. However, at the level of the individual sample member, the validity of symptom assessment by significant others could not be supported by data obtained from the physicians. With regard to the type of significant others interviewed, children reported more symptoms than partners. The use of significant others in after-death interviews can be a valid method with regard to the assessment of chronic diseases and symptoms on a group level. On an individual level this can be concluded only for chronic diseases with clearly observable consequences.