In view of increasing debate over possible benefit of more complete surgery compared to conservative procedures, a randomized controlled trial contrasting the then standard Halsted radical (RDL) operation with the more complete extended radical (EXT) mastectomy was initiated in 1973. Between November 1973 and July 1982, 123 women younger than 70 years of age and at clinical Stages I and II were enrolled. Of the total series, 112 were treated by the same surgeon and confirmed pathologically as having invasive mammary carcinoma. In this more homogeneous subgroup, the 10-year survival rates (and standard errors) were for RDL, 60% (+/- 7%) and for EXT, 74% (+/- 6%) (P value for comparison of survival curves = 0.13). In patients from this subgroup with central-medial tumors, comprising 62% of the total, survival after RDL at 10 years was 60% (+/- 8%), and after EXT 86% (+/- 6%) (P = 0.025). In the remaining patients with lateral tumors, survival rates were unaffected by treatment: 58% (+/- 13%) and 56% (+/- 11%), respectively (P = 0.62). Comparison of a nonrandomized series of 266 RDL and 124 EXT patients treated between 1960 and 1978 found differences consistent with those of the randomized study, although not statistically significant.