The major purpose was to explore the use of written case scenarios to investigate factors associated with regional variations in referral for consultation. Data were collected in Nova Scotia, where extensive computer-based records of utilization of health services are kept, and the nine health care regions are in sufficient proximity to make visits and interviews feasible. Interviews with 9 x 25 = 225 randomly-selected family physicians, using scenarios analogous to ICD-9 codes, permitted testing of hypotheses about the effects of selected variables on referral rates for hypothetical cases. The family physicians' self-reported referral behavior for the written case scenarios was compared, region by region, with the actual referral of analogous cases by physicians in the same region, as determined from Nova Scotia Health Services and Insurance Commission records. The results indicated that written scenarios provide a useful tool for studies of such variables. Generally, the family physicians responded appropriately to the information that described the hypothetical patients ("case cues"). However, comparisons of hypothetical and actual referral rates indicated that appropriate information about the environment within which referral decisions take place ("environmental cues") may be needed to explain actual regional variations in referral rates.