OBJECTIVES--In an epidemic: to measure the incidence and risk of complications of influenza; to determine the effect of pre-existing disease on complications; to estimate vaccine uptake and efficacy. DESIGN--Case-control study. SETTING--Primary care: two group practices. SUBJECTS--342 of the 395 cases of clinically diagnosed influenza reported to the general practice surveillance of infectious diseases scheme of the Public Health Laboratory Service during the 1989 epidemic, and 342 age and sex matched controls. INTERVENTIONS--Examination of records. MAIN OUTCOME MEASURES--Documented recognised complications; hospital admission; previous vaccination. RESULTS--Of 15 recognised complications, bronchitis was the commonest (rate 190.1/1000 cases) and significantly commoner in cases (summary odds ratio 9.7) after adjusting for higher consultation rates (mean 6.1 per annum v 4.2 among controls; p < 0.0001). No deaths were recorded. The risk of bronchitis complicating influenza was higher in patients with pre-existing illnesses regarded as an indication for vaccination (odds ratio 3.3; p < 0.0001). Observed vaccination efficacy in those with pre-existing illnesses and in elderly subjects was high (63% and 77% respectively) but uptake was low (4.5% and 6.1% respectively). CONCLUSIONS--Bronchitis complicates about one fifth of all cases of influenza presenting to general practitioners. Patients with pre-existing illnesses regarded as an indication for vaccination are particularly at risk. Vaccine uptake is extremely low, precluding an unequivocal demonstration of a protective effect.