Little is known about the accuracy of data reported in practice based primary care research. The Ambulatory Sentinel Practice Network (ASPN) undertook a 100% audit of 226 patients included in a study of spontaneous abortion (SAB). The audit was conducted to assess the feasibility of conducting audits in primary care research networks dispersed over large geographic areas, verify that patients met inclusion criteria, and assess the frequency of reporting errors using the medical record as a standard. Of the originally reported SABs, 24% could not be verified. The overall error rate was 4.5%, a total of 106 errors out of a possible 2,361. Seventy percent of these errors came from five of the 34 participating practices. Sixty-six percent of the records were error-free. Seventy-seven percent of the errors were associated with problems with methods and clustered into three categories: gravidity, gestational age, and dilation and curettage (D&C). According to this audit, the data reported by the practices for research purposes were very similar to the data found in the medical record.