OBJECTIVE: To assess whether scores on the three major quality-of-life questionnaires in oncology (FACT-G, FLIC, and EORTC QLQ-C30) are associated with modes of administration in a realistic clinical research setting. STUDY DESIGN AND SETTING: A heterogeneous sample of 1,265 cancer patients was recruited in Singapore. About one-fourth of the patients chose to have the interview administered by research staff; the rest self-completed the questionnaires. Multiple regression was used to adjust for differences in demographic and clinical characteristics between patients. An equivalence margin was defined as 0.25 standard deviations. RESULTS: Apart from one exception (the EORTC QLQ-C30 global functioning scale), all scales showed higher mean values in patients who were interviewed than patients who self-administered the questionnaires. For the physical and functional well-being scales of FACT-G and the physical and social functioning scales of EORTC QLQ-C30, the differences were small and the confidence intervals fell totally within the equivalence zone. The emotional well-being score of the FACT-G was different across modes of administration and the confidence interval fell outside the equivalence zones. There was no interaction between modes of administration and respondents' education level. CONCLUSION: The physical aspect of quality-of-life is not sensitive to interviewer administration but the psychological aspect is. Statistical adjustment for some scales is recommended.