To explore the feasibility of linking records to study health services and health outcomes for primary care patients.A cohort of patients from the Family Medicine Centre at Mount Sinai Hospital was assembled from the clinic's billing records. Their health numbers were linked to the Ontario Hospital Discharge Database. The pattern of hospital admission rates was investigated using International Classification of Diseases (ICD) codes for primary discharge diagnosis. A pilot case-control study of risk factor management for stroke was nested in the cohort.Family medicine clinic based in a teaching hospital.A cohort of 19,654 Family Medicine Centre patients seen at least once since 1991.Admission rates by age, sex, and diagnosis. Numbers of admissions for individual patients, time to readmission, and length of stay. Odds ratios for admission for cerebrovascular disease.The 19,654 patients in the cohort had 14,299 discharges from Ontario hospitals in the 4 years from 1992 to 1995, including 3832 discharges following childbirth. Some patients had many discharges: 4816 people accounted for the 10,467 admissions excluding childbirth. Excluding transfers between institutions, there were 4975 readmissions to hospital during the 4 years, 1392 (28%) of them within 28 days of previous discharge. Admissions for mental disorders accounted for the greatest number of days in hospital. The pilot study of risk factor management suggested that acetylsalicylic acid therapy might not be effective for elderly primary care patients with atrial fibrillation and that calcium channel blocker therapy might be less effective than other therapies for preventing cerebrovascular disease in hypertensive primary care patients.Record linkage combined with data collection by chart review or interview is a useful method for studying the effectiveness of medical care in Canada and might suggest interesting hypotheses for further investigation.