One hundred women with operable breast cancer were randomized to two equal groups. In group A the aim was to remove all fat tissue in the axilla. In group B suspectedly pathologic nodes were removed, and if no nodes were palpable the lower half of the axillary fat was excised. Irrespective of this grouping, partial mastectomy was generally performed for T1 tumour (17 in each group) and mastectomy for T2 tumour (33 in each group). Also irrespective of A and B grouping, radiotherapy including the axilla was given after partial mastectomy, and to women under 70 with T2 tumour and/or lymph-node secondaries. Arm volume and shoulder mobility were measured before and 3, 6 and 12 months after surgery. Arm oedema (greater than or equal to 10% volume increase) was found in 14 group A patients, but none in group B. Twelve of the 14 patients with arm oedema had been irradiated. Subjective sensation of arm swelling in the absence of an objectively measurable increase in arm volume was also more common after axillary dissection plus irradiation than after the other forms of treatment. Impairment of shoulder mobility was more frequent after axillary irradiation.