BACKGROUND: Heat-wave frequency, intensity, and duration are increasing with global climate change. The association between heat and mortality in the elderly is well documented, but less is known regarding associations with hospital admissions. OBJECTIVES: Our goal was to determine associations between moderate and extreme heat, heat waves, and hospital admissions for nonaccidental causes among Medicare beneficiaries >/= 65 years of age in 114 cities across five U.S. climate zones. METHODS: We used Medicare inpatient billing records and city-specific data on temperature, humidity, and ozone from 1992 through 2006 in a time-stratified case-crossover design to estimate the association between hospitalization and moderate [90th percentile of apparent temperature (AT)] and extreme (99th percentile of AT) heat and heat waves (AT above the 95th percentile over 2-8 days). In sensitivity analyses, we additionally considered confounding by ozone and holidays, different temperature metrics, and alternate models of the exposure-response relationship. RESULTS: Associations between moderate heat and hospital admissions were minimal, but extreme heat was associated with a 3% (95% CI: 2%, 4%) increase in all-cause hospital admissions over the subsequent 8 days. In cause-specific analyses, extreme heat was associated with increased hospitalizations for renal (15%; 95% CI: 9%, 21%) and respiratory (4%; 95% CI: 2%, 7%) diseases, but not for cardiovascular diseases. An added heat-wave effect was observed for renal and respiratory admissions. CONCLUSION: Extreme heat is associated with increased hospital admissions, particularly for renal causes, among the elderly in the United States.