To assess the magnitude of fatal Rocky Mountain spotted fever (RMSF) and to evaluate the completeness of national surveillance for this occurrence from 1983 through 1998, two independent sources of RMSF mortality data were analyzed using a capture-recapture method. Two hundred twenty-four deaths reported through RMSF case report forms (CRFs) were compared with 304 RMSF-associated deaths recorded in the United States multiple cause-of-death (MCD) database. Demographic, geographic, seasonal, and temporal characteristics of decedents were remarkably similar between sources. Median annual deaths ascertained from CRF and MCD data sources were 11 (range= 5-35) and 18 (range= 5-39), respectively. Decedents were matched between sources by year, state, age, sex, race, and month-of-death; 111 deaths were matched to both sources, and percent concordance between CRFs and MCD data during the 16-year study period was 27%(range= 7-45%). An estimated 612 RMSF-associated deaths occurred during the study period (median= 37 per year, range= 16-64), suggesting that approximately 400 fatal cases of RMSF went unreported to national surveillance during the period 1983-1998, for an estimated completeness of CRF reporting of 36%.