Until the recent introduction of imatinib, interferon alfa was the standard treatment for patients in the chronic phase of chronic myeloid leukaemia. We compared survival of 143 such patients, who did not respond to interferon alfa and were treated with imatinib, with that of 246 historical controls who received conventional treatment. Patients on imatinib showed an overall survival advantage (relative risk 0.54, 95% CI 0.31-0.93). However, although patients on imatinib who achieved at least some degree of cytogenetic response after 6 months had better survival than controls (0.13, 0.05-0.39), those with no cytogenetic response to imatinib had significantly worse survival (1.69, 1.09-2.64). Our findings suggest that cytogenetic responders obtain benefit from imatinib but patients who show no cytogenetic response should be given alternative treatment without delay. We confirmed these results in a case-matched analysis.