OBJECTIVE--To determine the prognostic significance of altered plasma viscosity and erythrocyte aggregation in unstable angina. DESIGN--A prospective study of 96 consecutive patients with unstable angina allocated to one of two groups according to predefined threshold values for plasma viscosity and erythrocyte aggregation at study entry. The patients received a standardised treatment and were followed up for six months or until angioplasty or bypass surgery. MAIN OUTCOME MEASURE--Frequency of myocardial infarction. RESULTS--Myocardial infarctions occurred in 7/26 patients with a plasma viscosity greater than or equal to 1.38 mPa s and in 8/35 with a rate constant of erythrocyte aggregate formation greater than or equal to 0.5 mPa (corrected for plasma viscosity) but in only 4/70 with a plasma viscosity less than 1.38 mPa s and in 3/61 with an erythrocyte aggregation less than 0.5 mPa (odds ratios: 6.1 (95% confidence interval 1.3 to 31), p = 0.008, and 5.7 (95% CI 1.2 to 35), p = 0.016). Plasma viscosity and erythrocyte aggregation were more predictive of myocardial infarction than age, male gender, fibrinogen concentration, ST segment abnormalities, or coronary score. Furthermore, Holter monitoring with ST segment analysis showed that ischaemic episodes were more common in patients in whom the rate constant of erythrocyte aggregate formation was greater than 0.5 mPa (15/27 v 17/50, p = 0.029). Cardiac troponin T release was increased in patients with a plasma viscosity of greater than 1.38 mPa s (10/26 v 9/70, p = 0.010). CONCLUSIONS--In patients with unstable angina a considerable increase in plasma viscosity and erythrocyte aggregation identified a subgroup of patients at a high risk of acute myocardial infarction in whom medical treatment was likely to be unsuccessful.