Preliminary axillary node sampling with intraoperative reporting of imprint cytology to detect positive nodes was assessed in 50 consecutive patients with T1 or T2 breast cancer undergoing axillary clearance, 21 of whom were histologically node positive. A mean of 3.5 (range 1-4) nodes were sampled. Axillary clearance yielded a mean of 11.7 (range 4-32) nodes. Of node-positive patients, sampled nodes detected 18 of 21 on immediate processing, one on subsequent paraffin sections, and failed to detect two (one of ten and one of 17 positive nodes). If the technique had been used during operation to select patients for axillary clearance, 29 of 50 would have avoided the operation, one of two node-positive patients would have required delayed axillary treatment and two of 21 with minimal axillary disease would not have received further axillary treatment or adjuvant therapy. The technique may be useful in reducing the morbidity of breast cancer surgery without increasing the risk of locoregional or distant recurrence.