In 1994, the Italian and the German Chronic myeloid leukemia (CML) trials comparing interferon-alpha (IFN-alpha) with conventional chemotherapy were published. The survival advantage in favor of IFN-alpha compared with hydroxyurea (HU; 72 v 52 months) was significant in the Italian (P < .002), but not in the German trial (66 v 56 months, P < .44). We set up a collaborative study to identify the reasons for the different outcomes. There are major differences in the trial protocols concerning admission criteria, treatment strategy, and definitions. The German patients were older and more seriously sick. Fifty-two of the 327 patients in the German IFN and HU arms did not fulfil Italian admission criteria, and 41 of the 322 Italian patients did not fulfil German admission criteria. Using mutually uniform admission criteria, the median survival times of the IFN patients are 76 (Italian) and 72 (German) months (P = .56). The Italian group administered IFN combined with HU as needed, whereas the German group strictly used IFN as monotherapy with rerandomization to busulfan (BU) or HU after IFN resistance or intolerability. The differences seen between the Italian and the German trial results can be accounted for by objective differences in study design, especially the admission criteria, treatment strategy, and bias due to intention to treat analysis. The detailed analysis of the data suggests that the combination of IFN with HU as needed is more effective than either agent alone.