The correct approach to the management of the asymptomatic carrier with a recognized inherited thrombophilic disorder is uncertain because reliable in formation of the risk of spontaneous (unprovoked) throm bosis in these disorders is not available. To determine the best available estimate of the annual incidence of spon taneous thrombosis in asymptomatic carriers of disorders that have been linked to familial thrombophilia, we per formed a literature review. Using Medline search from 1965 to 1992, supplemented by manual searches, we re trieved all articles that presented data on antithrombin III, protein C, protein S, dysfibrinogenemia, plasmino gen, histidine-rich glycoprotein, heparin cofactor II, and fibrinolysis in relation to thrombosis. Publications were included in the analysis if they (1) reported one or more probands with thrombotic disease and a heterozygous biochemical abnormality of the hemostatic system, (2) assessed the presence of this abnormality in family mem bers independent of the presence or absence of a history of thrombotic disease, and (3) assessed the presence of a history of thrombotic disease in all available family mem bers. The biochemical status and clinical details of all family members reported were extracted from each eligi ble article. For each abnormality the odds ratio for throm bosis was compared in family members with and without the biochemical abnormality. If applicable, thrombosis- free survival and age-specific incidences of thrombosis were calculated. The thrombotic episodes were classified as spontaneous or secondary to a recognized risk factor, and the proportion of spontaneous episodes was calcu lated. The influence of diagnostic suspicion bias in symp tomatic patients with a family history of thrombosis was reduced by recalculating the absolute incidence of throm bosis from the odds ratio after adjusting the incidence of venous thrombosis in nonaffected family members to that observed in the general population. Statistically signifi cant associations between the presence of a biochemical abnormality and a history of venous thrombosis were found for antithrombin III deficiency types 1 and 2a and 2b, protein C deficiency type 1, and protein S deficiency type I. Dysfibronogenemia was statistically significantly associated with venous as well as arterial thrombosis. Thirty-five to 67% of the events were classified as being provoked, as they occurred following exposure to a rec ognized risk factor for thrombosis. The recalculated an nual incidence of spontaneous thrombosis was 0.6 to 1.6%/year. It is concluded that this relatively low inci dence does not warrant life-long continuous use of anti coagulant prophylaxis since the reported risk of major and fatal bleeding associated with the use of oral antico agulants is 2-3 and 0.4%/year, respectively.