The aim of this study was to investigate and compare the prognosis after treatment for loco-regional recurrences (LR) after (modified) radical mastectomy (MRM) or breast conserving therapy (BCT), in terms of overall survival and time to subsequent LR, in patients originally treated in two European randomised trials. In EORTC trial 10801 and DBCG trial 82-TM, 1,807 patients with stage I and II breast cancer were randomised to receive MRM or BCT from 1980 to 1989. All patients with a LR in these trials were analysed for survival and time to subsequent LR after salvage treatment. Of these, 133 patients had their LR as a first event, the majority within 5 years after initial treatment. The prognostic significance for survival and time to subsequent LR after salvage treatment was analysed in uni-, and multivariate analyses for a number of original tumour- and recurrence-related variables. After salvage treatment of LR after MRM or BCT, actuarial survival curves and the actuarial locoregional control curves were similar. The 5-year survival rates were 58% and 59% and the 5-year subsequent loco-regional control rates 62% and 63%, respectively. In a multivariate analysis, pN category (P = 0.03), pT category (P = 0.01) and vascular invasion (P = 0.02) of the primary tumour were the only independent prognostic factors for survival, whereas extensive LR (P < 0.001), interval < or = 2 years (P < 0.002) and pN+ at primary treatment (P = 0.004) were significant predictive factors for time to subsequent LR. The type of original treatment (MRM or BCT) did not have any prognostic impact. It is concluded that the survival and time to subsequent LR after treatment for an early loco-regional recurrence after MRM or BCT was similar in these two European randomised trials. This suggests that both after MRM and BCT an early LR is an indicator of a biologically aggressive tumour; early loco-regional relapse carries a poor prognosis and salvage treatment only cures a limited number of patients, whether treated by MRM or BCT originally.