Objective The role of secondary cytoreductive surgery (SCR) in platinum-sensitive recurrent ovarian cancer (ROC) remains controversial. The overall survival (OS) benefits for surgery reported in observational studies may be due to the selection of patients with better prognosis. Methods Using data from the CALYPSO trial, OS of patients who had SCR was compared to those treated with chemotherapy alone. Multivariate analyses were performed to adjust for prognostic factors. We also tested for an interaction between baseline prognostic groupings and the benefit of surgery. Results Of the 975 patients randomised in CALYPSO, 19% had SCR and 80% had chemotherapy alone. OS was longer for the SCR group than for chemotherapy alone (median, 49.9 vs. 29.7 months; adjusted hazard ratio (HR), 0.68; P = 0.004). For patients with SCR, the 3-year OS was 72% for those with no measurable disease, and 28% if residual tumour was larger than 5 cm. Patients with good prognostic features benefited the most from SCR (HR 0.43; P < 0.001). The benefit of SCR was less in patients with poorer prognostic features (test of trend P < 0.001). Conclusion SCR was associated with improved OS in platinum-sensitive ROC, particularly in patients with favourable prognostic characteristics. However, these findings may be due to selection bias, and hence randomised trials are still essential. [ABSTRACT FROM AUTHOR]