Objective: To analyse congruence on medication throughout patient courses, including an acute admission to a medical department. Design: A prospective, observational study. Data were collected from patient records in primary health care, hospital departments, from the Health Insurance database and through patient interviews. Setting: Departments of internal medicine, general practice and patients' homes. Main outcome measures: Number, type and character of discrepancies between paired sources of information on patient medication at predefined time points throughout the complete patient course. Assessment of likelihood and severity of potential untoward effects of discrepancies. Results: Data were obtained for 75 of 99 consecutive patients included. Patients stated use of four drugs (median, range 0-17) at admission, five (0-16) at discharge and four (0-15) 1 month after discharge. At admission, 11 patients used no drugs. A median of one (0-20) to three (0-16) discrepancies per patient were identified in seven paired source comparisons with no improvement along patient course. Full agreement throughout the course was found in six patients (8%; 95% confidence interval: 3-17%). No association was found among source discrepancies and number of drugs and age. Of discrepancies, 4-13% were considered serious and likely to cause untoward effects. Discrepancies due to synonymous and analogous drugs accounted for 2-7% of the discrepancies. Conclusion: Congruence between sources of information on medication throughout patient courses cannot be obtained with separate medication charts. Discrepancies among patient, general practitioner and hospital give rise to a definitive risk of serious untoward effects.