There are few population-based studies of the natural history and epidemiology of herpes zoster. Although a relatively common cause of morbidity, especially among the elderly, contemporary estimates of herpes zoster incidence are lacking. Herein we describe a population-based investigation of incident and recurrent herpes zoster from 1990 through 1992 in a health maintenance organization. The health maintenance organization's automated medical records contain clinical and administrative information about care rendered to patients in ambulatory settings, emergency departments, and hospitals. Cases of herpes zoster were ascertained by screening the medical record for coded diagnoses. The predictive value of a herpes zoster diagnosis code was determined by review of a sample of patient records. Records from all patients with potential recurrences were also reviewed. The overall incidence, based on 1075 cases in 500,408 person-years, was 215 per 100,000 person-years (95% confidence interval, 192 to 240 per 100,000) and did not vary by gender. Although the rate increased sharply with age, approximately 5% of the cases occurred among children younger than 15 years. Infection with human immunodeficiency virus was documented in 5% of the persons with incident herpes zoster and cancer in 6%. Four persons had confirmed recurrences of herpes zoster (744 per 100,000 person-years; 95% confidence interval, 203 to 1907); three of these persons were infected with the human immunodeficiency virus. The recorded incidence of herpes zoster was 64% higher than that reported 30 years ago; the age-standardized rate was more than twofold higher. Immunosuppressive conditions had little impact on overall incidence, although they were strongly associated with early recurrences.