To assess the role of routine follow-up in current management of breast cancer. Retrospective review. Teaching hospital, The Netherlands. 270 patients who presented with recurrent breast cancer, 1974-90. Recurrence was coded as asymptomatic or symptomatic and related to survival. 170 (63%) of the recurrences were detected when they were symptomatic and 100 (37%) when they were not. The groups differed significantly according to the site of recurrence; 45/100 recurrences were local in the asymptomatic group compared with 23/170 (14%) in the symptomatic group. There was no significant difference in disease-free survival between the two groups. Overall 5-year survival after primary treatment for all recurrences (locoregional and distant) was significantly better (p=0.0003) in the asymptomatic group (62/100) than in the symptomatic group 79/170 (46%). However, when locoregional and distant recurrences were analysed separately no significant differences were found between both groups in overall survival after primary treatment or survival after detection of recurrence. The 5-year overall survival after primary treatment for distant recurrence was 26/47 (55%) in the asymptomatic group compared with 62/134 (46%) in the symptomatic group (p=0.13). For locoregional recurrence these figures were 35/45 (78%) and 14/23 (61%), respectively (p=0.34). Routine follow-up hardly affected the course of locoregional recurrence. Only five of 75 patients with local recurrence (7%) developed uncontrolled local disease, 2 of whom were initially detected during routine follow-up. We conclude that in the current management of breast cancer the medical impact of follow-up is low, so follow-up visits after treatment for breast cancer are hardly warranted.