BACKGROUND: The availability of a vaccine for the prevention of herpes zoster has increased interest in methods to measure zoster disease burden. Hospitalizations assigned a zoster diagnosis code have been used as indicators of severe zoster in prior studies. However, a zoster diagnosis code may not be a specific indicator of severe zoster illness, because the code may be assigned to a hospitalization for another cause in a person with coincident zoster. METHODS: To assess the validity of a hospital diagnosis code of zoster as an indicator of hospitalizations that are attributable to zoster, we identified all hospitalizations with a zoster diagnosis code assigned in any position among members of a managed-care organization who were >or=50 years of age during 1992-2004. Of those, we selected a sample of 260 hospitalizations for chart review. RESULTS: Chart reviews were completed for 225 hospitalizations. Sixty-five (29%) were because of zoster or a complication of zoster treatment, and an additional 9 (4%) were because of postherpetic neuralgia or a complication of postherpetic neuralgia treatment. Although the overall age-adjusted rate of hospitalizations with a zoster diagnosis code was 42.5 hospitalizations per 100,000 population per year, the estimated rate of hospitalizations because of zoster, postherpetic neuralgia, or adverse effects of a medication used to treat zoster or postherpetic neuralgia was only 14.0 hospitalizations per 100,000 population per year. CONCLUSIONS: Rates of hospitalizations associated with a zoster diagnosis code will substantially overestimate the burden of hospitalizations attributable to zoster in older adults.