This paper investigates the influence on patient survival of an individual clinician's experience with a particular treatment regimen and the collective experience of the centre through which the patient is recruited to a particular clinical trial. Data were available from two series of randomized trials, one series in small cell and the other in non-small cell lung cancer, which were conducted by the Medical Research Council Lung Cancer Working Party. Successive small cell trials used the same chemotherapy regimen and successive non-small cell trials the same radiotherapy regimen. We found no evidence that either the degree of experience of individual clinicians with the regimen or the participation in terms of number of patients recruited to centres influenced patient survival. There is a strong suggestion, in non-small cell lung cancer that, as clinicians' experience with the regimen extended, they became increasingly likely to admit patients in poor condition to the trials. We stress how important it is to take account of patient prognostic characteristics in such an analysis, as we found that unadjusted comparisons suggested negative influences of increasing experience with the regimen. The implications for routine audit are self evident, in particular that it is not a substitute for the randomized controlled trial.