OBJECTIVE: The pharmacy's recorded days supply variable is frequently used in outcomes research to estimate a prescription's intended period of use, but little is published on factors that might influence its validity. We examined the effect of a change in copayment structure on the accuracy of the days supply field of a prescription records database. METHODS: Prior to January 1, 2002, Kaiser Permanente Northwest enrollees were able to obtain a 90-day supply of medication for one copay. After this date, a copay was required for each 30-day medication supply. We collected dispensing data for oral antidiabetic medications from January 1, 1996-December 31, 1997 (90-day copay), and from January 1, 2002-December 31, 2002 (30-day copay). Accuracy of the recorded days supply variable was compared to a gold standard (calculated by pharmacist (KS) interpreting each prescription's directions for use. RESULTS: There were 3155 and 1495 records in the 90-day copay and 30-day copay periods, respectively. The mean difference (±SD) between gold standard and recorded days supply under the 90-day copay was 0.92 days (±24.3); 72.8% were in complete agreement, 10.7% were +1 to +30 days different, 5.2% were -1 to -30 days different, maximum range of -200 to 370 days different. The mean difference (±SD) under the 30-day copay was -0.33 (±10.8) days; 88.3% were in complete agreement, 9.1% were +1 to +30 days different, and 5.1% were -1 to -30 days different, range of –90 to + 90 days different. CONCLUSION: After requiring additional copayments for a 90-day supply, the pharmacy's recorded days supply was more likely to be in agreement with the directions for use. Both pharmacy staff and enrollees may have contributed to the improvement by verifying that copays were accurate. When both provider and enrollee have a financial interest, prescription days supply data may be more accurate.