As part of a phase II study of the EORTC Soft Tissue and Bone Sarcoma Group, 15 centres took part in a programme to evaluate the quality of treatment delivered and data collected. The centres were visited and facilities for treatment and data management were reviewed. Source data in randomly selected patient hospital records were compared with information which had previously been completed on case record forms and returned to the EORTC Data Centre. The review included 71% of the patients entered into the study and 76% of the treatment cycles. Chemotherapy was prepared by nurses or clinicians in 58% and by pharmacists in 42% of the centres and was administered by specialist nurses in 67% of the hospitals. 8776 items were checked with source data. 3.4% were incorrect, 0.2% were missing and 30% could not be verified as correct (mainly related to the lack of recording of toxicity data in hospital records). The mean doses of chemotherapy delivered and treatment intervals were those stipulated in the protocol but 21% of the cycles were delayed for avoidable reasons. Several modifications to the procedures for running chemotherapy trials were suggested by this survey including the use of a systematic checklist for recording toxicity and chemotherapy administration and the development of quality assurance programmes in other collaborative groups and single centres to ensure that published results are credible.