The sensitivity of passive reporting of Guillain-Barré syndrome (GBS) to the Vermont Department of Health from 1980 to 1985 was compared to that of computerized hospital discharge abstract data. Written hospital discharge summaries were reviewed for clinical data to validate the computerized abstracts. In all, 51 definite and probable cases of GBS were identified from hospital data during a period when only 4 cases (8%) had been reported to the health department through passive physician reporting. Based on the hospital data, the incidence of this syndrome in Vermont was 1.6/100,000 population/year. The incidence rate for males was 1.5 times that for females. No geographical or seasonal clustering of cases was found. These epidemiological features are consistent with previously published data on the syndrome and suggest that the incidence has not changed significantly in the past 10 years. Incidence rates for GBS based on passively reported cases markedly underestimate the true incidence rate. Although limited by the lack of timeliness for public health surveillance, computerized hospital discharge data are readily available in many states and may be more sensitive in detecting cases, compared to passive surveillance. They may be a useful tool for establishing baseline rates and examining long-term trends for selected acute diseases like GBS for which there are well-established diagnostic criteria and that usually result in hospitalization.