Unexplained differences in the density of in-patient management in one federal state in Germany led to a regional survey of physicians in independent and hospital practice, which aimed to describe more precisely the determinants of referral and admission behaviour. Brief typical case descriptions (vignettes) were designed, as an instrument of data collection, dealing with two examples of management problems: upper and lower abdominal pain. The urgency of inpatient treatment was ranked using clinical scores and guidelines. Social characteristics for the patient (age, gender, social situation, preference for/against hospitalisation, day of the week on which the patient presented) were randomly assigned to the case vignettes. Each physician was asked by mail to decide on the management of 10 upper and 10 lower abdominal vignettes each. The physicians were also asked to provide additional information on the characteristics of their practice or hospital. The data were analysed using multivariate hierarchical models. A 28 % response rate meant that vignettes were available from 455 general practitioners and internal medicine specialists in independent practice, as well as 261 hospital surgeons and internal medicine specialists, together with responses from 31 physicians from the medical service of the German statutory health insurance (MDK). 7376 upper abdominal and 7335 lower abdominal vignettes were analysed. Admission rates reflected the graded severity of the symptoms built into the vignettes. Hospital physicians wanted to admit the vignette patients much more frequently than physicians in independent practice wished to refer them. Older patients, independent of symptoms, were more frequently referred or admitted than younger patients. In the case of acute symptoms it is the day of the week when the patient consulted the physician and in the case of elective surgery the patient's preference that are important for hospitalisation. The results show that medical decisions on case management are made using reasonable problem-specific preferences. The probability of actually instigating admission was, other things being equal, much higher in hospital physicians than in office-based physicians. As an instrument of comparative research into medical care, case vignettes have practical advantages in relation to medical audits and standardised patients. They can also be used for teaching, examining, documenting and quality assurance purposes.