This paper describes the development and testing of the Patient Registration Form (PRF), a standardized instrument for the description of patients seen by consultation-liaison (C-L) psychiatrists and psychosomatists in general hospitals, the referral patterns, the C-L interventions and their outcomes. The PRF study is part of a large multi-centre. European investigation on the effectiveness of mental health service delivery, conducted by the European C-L Workgroup for General Hospital Psychiatry and Psychosomatics (ECLW) and performed in the framework of the of the E.C. 4th Medical and Health Research Program. The final version of the PRF consists of 68 items. It was developed by the Program Management Group (PMG) and National Coordinators (NC) after long preparatory studies to assure both face and content validity and pilot testing. Two hundred and twenty consultants, who required 40 hours of training and came from 14 different European countries and 90 different sites, participated in the final reliability study. The PRF was tested in 13 written case histories. A "gold standard' for the correct answers in each item was decided by "consensus ratings' of the PMG and a subsequent 80% agreement by the NCs. A high standard (average kappa (k) > or = 0.70; at least 2/3 of the PRF items, kappa > or = 0.70) was required for the rater to be considered as "reliable' (RR). The consultants considered the PRF both "feasible' and 93.2% of them fulfilled the RR criteria. The calculated rater-"gold standard' reliability was satisfactory: only four PRF items were identified with low agreement coefficients and no biases were observed cross-nationally in the ratings. Given the implications of misclassification for evaluating C-L effectiveness and services, these results are important and the achievement unprecedented.