This study examines the validity of self-reported hospital admission data obtained by telephone interview from a sample of 563 participants in a follow-up study in Auckland, New Zealand. The personal recall of hospital admissions over a 4-year period between 1988 and 1991 was compared with a national computerized record of all hospital admissions. Only two admissions were recalled by participants but not recorded on the hospital record. However, the study found significant underreporting of admissions by participants. Of those participants who could be contacted, 58% recalled all their admissions recorded on the computerized database. Additionally, 26% recalled some of the admissions, and 16% recalled none of the admissions. The age, sex, and social class status of the participants did not appear to significantly influence recall ability. A greater proportion of "first" admissions (68%) was recalled than that of "readmission" (45%). Follow-up studies that rely on similar methods for collecting self-reported hospital admission data may significantly underestimate admission rates, particularly those of readmissions for the same condition.