A number of possible risk factors, clinical, radiological, biochemical, and social have been examined in one group of 34 cases of mitral valve disease and four of lone atrial fibrillation, all of whom gave a history of embolism, and also in a group of 24 cases of mitral valve disease who gave no such history. All were on long-term anticoagulant treatment. Comparison of the two groups disclosed no features that would distinguish those who ran the greater risk of embolism from the remainder other than that the men in the former group were older and of shorter stature. At comparable age, the women with a history of embolism had smaller left atria. Examination of a special group of nine patients who had had embolism despite anticoagulant treatment also disclosed no special features, other than a greater age at the time of their first embolism. Fibrinolytic activity was less and the level of beta thromboglobulin was greater than normal in both groups. Those with proven risk of embolism were not distinguished by higher values of this platelet protein. Neither variable correlated with the type of heart disease. No positive association was found between cigarette smoking and the risk of embolism. Embolism was significantly more likely to occur between 0801 and 1600 hours than at other times of day.