The Caerphilly and Speedwell studies have previously reported the predictive power of heat-precipitation, nephelometric, fibrinogen for 10-year incidence of ischaemic heart disease. A Clauss, clotting time, fibrinogen was also measured at baseline, but has not previously been reported. The predictive power of the two assays is compared. Both methods were employed on fasting blood samples from a total of 4391 men aged 45-63 years. Over the following 10 years 533 (12%) developed major ischaemic heart disease. Nephelometric fibrinogen was higher by 0.33 g/l among the men who developed disease; clottable fibrinogen was higher by 0.20 g/l. This difference is statistically significant (P=0.01). Relative odds of developing heart disease increased steadily to 3.53 (P<0.0001) in the 20% of men with the highest nephelometric fibrinogen; for clottable fibrinogen the corresponding relative odds increased to 2.24 (P<0.0001). When both measures of fibrinogen were included in logistic regression models together with age and smoking habit, the trend for incidence to increase with increasing nephelometric fibrinogen remained highly significant (P<0.0001), whereas for the Clauss fibrinogen the trend almost entirely disappeared (P = 0.37). We conclude that functional assays of clottable fibrinogen may not reflect all of the mechanisms which mediate the association between fibrinogen and cardiovascular disease and that assays of both 'heat precipitable' and 'clottable' fibrinogen should be included in all future prospective studies.