The association between nonsteroidal anti-inflammatory drugs (NSAIDs) and neutropenia is based primarily on case reports only. A population-based, case-control study was performed with Medicaid claims data from six states. Cases were defined as patients hospitalized with neutropenia. Four controls per case were randomly chosen, matched for age, sex, state, and year. The frequency of exposure to NSAIDs in the 30 days before hospital admission in the cases was compared with the frequency in the identical period in the controls. The diagnosis of neutropenia was validated by review of medical records. The crude odds ratio for NSAIDs as a class was 3.3 (90% confidence interval, 1.6 to 6.6). The multivariate adjusted odds ratio was 4.2 (90% confidence interval, 2.0 to 8.7). No single class of NSAID, nor any individual NSAID, was associated with a unique risk, although the data on individual NSAIDs were sparse. Even excluding phenylbutazone and indomethacin, an increased risk was observed (3.5 [1.6 to 7.6]). Neutropenia is associated with the use of NSAIDs. However, given the low incidence of this disease, the additional number of cases of neutropenia caused by the use of NSAIDs is small. These data do not support the existence of a risk restricted to selected NSAIDs only.