Physicians can potentially play an important role in the early detection of cancer. Interventions designed to encourage these activities have been shown to improve physician performance for up to 1 year. To assess their real value, improved physician performance must be judged over the longer term. The Cancer Prevention in Community Practice Project assisted a random subset of practices in implementing cancer early detection office systems. One year later, these practices were found to have provided more indicated breast and colorectal cancer early detection services than practices that did not receive assistance. This report addresses whether 12-month improvements in breast and colorectal cancer early detection were durable at 24 months despite no appreciable ongoing project support. A cross-sectional survey of 20-30 established patients/practice was conducted 24 months after the introduction of the intervention. These results were compared with base-line, 6-, and 12-month cross-sectional surveys to determine whether increases in indicated services and recommendations persisted. A longitudinal model for binomial data was used to quantitatively assess durability of effects. Ninety-nine practices participated, and 81 provided data at all 4 evaluation intervals. In office systems practices, improvements in stool occult blood testing and self breast examination recommendations to patients were maintained between 12 and 24 months while improvements in mammography recommendations and clinical breast examinations declined somewhat but remained superior to performance in control practices. Some improvements in physician early detection of cancer performance were maintained between 12 and 24 months. Future studies of physician behavior change should include follow-up beyond 12 months.