We studied anxiety levels in 68 patients who had been randomized to adjuvant chemotherapy v observation on two Eastern Cooperative Oncology Group (ECOG) protocols. All subjects were women who had undergone total or modified radical mastectomy for breast cancer. Immediately before breast protocol randomization and again 3, 6, and 12 months later, patients completed two self-report measures: the State-Trait Anxiety Inventory and the SCL-90. There were no consistent trends in anxiety levels over time. At each test point, patients under observation displayed higher anxiety scores than did patients receiving adjuvant therapy, but these differences failed to attain statistical significance. Power calculations indicate that these results rule out the possibility of major differences in anxiety levels among patients randomized to observation v adjuvant therapy, but a larger patient sample is required before a definitive statement can be made about smaller differences.