OBJECTIVES: Economic evaluations conducted from a societal perspective may include estimates of out-of-pocket expenses. Surveys and administrative registers offer two sources of out-of-pocket cost data for use in economic evaluations. We highlight potential problems in using questionnaires or administrative data alone. METHODS: Data are obtained by surveys and from an administrative register. Survey data include responses concerning out-of-pocket expenditure on prescription medications from men (N @1,400) who have been diagnosed as having prostate cancer prior to the 2004 survey. Register data concerning out-of-pocket expenditure on physician-prescribed medications are available for the survey population for pertinent time periods from the Finnish Social Insurance Institution. Register data and questionnaire data are linked using unique personal identification codes. Statistical comparison of these sources of estimates is performed using t-tests. RESULTS: The survey–based estimate of mean out–of–pocket expenses in the previous 12 months is €297, the corresponding figure from the administrative register is €259. The listwise estimate of survey–based overestimation is €38 and the difference between the two estimates of mean out–of–pocket expenses are statistically significant at the 5% level. We find that, performing imputation for unit and item non–response in the survey database and for non–inclusion in the administrative register, the survey–based estimate becomes 252€ and the register–based estimate becomes €255 — no longer a statistically significant difference. CONCLUSIONS: The problems of recall error and non–response can complicate the estimation of out–of–pocket expenditure using survey methodology. High–quality public sector administrative registers can serve the information needs of health economic evaluations. However, register–based information can also be usefully supplemented by questionnaire responses and vice versa.