Doubts are expressed on the validity of patient satisfaction questionnaires. High satisfaction levels are consistently reported. Within the European Organisation for Research and Treatment of Cancer (EORTC) quality of life group, we developed a cancer inpatient satisfaction questionnaire (QLQ-SAT32), adopting several precautions to overcome the ceiling effect commonly reported in satisfaction ratings. Since patients are often more critical when expressing themselves in an interview, in order to study the validity of the QLQ-SAT32, we assessed the agreement between self- and interview-administered QLQ-SAT32 responses. One hundred and twenty three patients were asked to complete the QLQ-SAT32 at home within 2 weeks of hospital discharge and were randomly allocated to participate in a telephone interview-administration of the QLQ-SAT32, either before or after self-completing the QLQ-SAT32. One hundred and four of them completed both modalities of questionnaire administration. Correlation and agreement between self- and interview-administered QLQ-SAT32 ratings were examined, for each subscale and the general satisfaction item of the QLQ-SAT32, using Spearman correlation, intra-class correlation coefficients (ICC) or weighted kappa coefficients. Agreement showed excellent for the doctors' and nurses' subscales and satisfactory for the services' subscale and the general satisfaction single item. The telephone interview-administration modality did not prove an adequate procedure to assess the validity of a patient satisfaction questionnaire.