Accurate measurement of change in tumour size is a prerequisite for the use of response-based regimens of primary systemic therapy for breast cancer. This study evaluated the accuracy of clinical assessment, mammography and ultrasonography in measuring tumour size and in monitoring response to treatment. Size was determined during the week preceding surgery and actual size measured from resected specimens. Sequential measurements were performed in 35 patients undergoing primary systemic treatment. There was moderate correlation between pathological and clinical size (n = 51, r2 = 0.68, P < 0.0001). Close correlation with pathological tumour size was observed for mammographic (n = 45, r2 = 0.84, P < 0.0001) and ultrasonographic (n = 52, r2 = 0.89, P < 0.0001) tumour size. Response was correctly evaluated by clinical assessment in 31 of 35 patients, by mammography in 20 of 35 and by ultrasonography in 31 of 35. Actual tumour size can be measured accurately by available imaging techniques but ultrasonography is the most practical and accurate method for monitoring response.