PURPOSE: Five years of adjuvant tamoxifen therapy for estrogen receptor (ER) -positive breast cancer is more effective than 2 years of use. However, information on tamoxifen discontinuation is scanty. We sought to identify predictors of tamoxifen discontinuation among older women with breast cancer. PATIENTS AND METHODS: Within six health care delivery systems, we identified women >or= 65 years old diagnosed with stage I to IIB ER-positive or indeterminant breast cancer between 1990 and 1994 who had filled a prescription for adjuvant tamoxifen. We observed them for 5 years after initial tamoxifen prescription. We used automated pharmacy records to validate tamoxifen prescription information abstracted from medical records. The primary end point was tamoxifen discontinuation, operationalized as ever discontinuing tamoxifen during 5 years of follow-up. We used Cox proportional hazards to identify predictors of tamoxifen discontinuation. RESULTS: Of 961 women who were prescribed tamoxifen, 49% discontinued tamoxifen before the completion of 5 years. Discontinuers were more likely to be aged 75 to less than 80 years (v < 70 years; hazard ratio [HR] = 1.41; 95% CI, 1.06 to 1.87), be aged >or= 80 years (HR = 2.02; 95% CI, 1.53 to 2.66), have an increase in Charlson Comorbidity Index at 3 years from diagnosis (HR = 1.52; 95% CI, 1.18 to 1.95), have an increase in the number of cardiopulmonary comorbidities at 3 years (HR = 1.75; 95% CI, 1.34 to 2.28), have indeterminant ER status (v ER-positive status; HR = 1.36; 95% CI, 1.00 to 1.85), and have received breast-conserving surgery (BCS) without radiotherapy (v mastectomy; HR = 1.62; 95% CI, 1.18 to 2.22). CONCLUSION: Attention to nonadherence among older women at risk of discontinuation, particularly those receiving BCS without radiotherapy, might improve breast cancer outcomes for these women.