The presence of antiphospholipid antibodies (aPL) in 188 unselected patients with systemic lupus erythematosus was studied using the recalcification time, kaolin clotting time (KCT), dilute Russell's viper venom time (dRVVT) and anticardiolipin ELISA (aCL) to identify patients with a high or low risk of thrombosis among patients with aPL. aPL were detected by at least one method in 104 (55%) of the patients. Despite heterogeneity, lupus anticoagulant (LA) methods correlated reasonably well with each other (r = 0.736-0.968), but poorly with aCL (r = 0.241-0.549). Positivity in LA assays and immunoglobulin G (IgG)-aCL were associated with patients who experienced thrombosis (P less than 0.001 for all assays). Patients with both LA and aCL had experienced thrombosis more often than those having only one (odds = 6.3, P less than 0.001). When patients with aPL were ranked by relative strength of the finding and divided into tertiles, a history of thrombosis was associated with membership in the strongest tertile of at least one assay (odds = 4.2, P = 0.002). LA and aCL had similar sensitivities for thrombosis (61% and 63%, respectively), but LA was more specific than aCL (79% vs 53%). The best combination of two assays was KCT with dRVVT (61% sensitivity, 87% specificity). Maximal sensitivity (71%) for thrombosis could be achieved by adding IgG-aCL to these two assays, but specificity was lower (73%). In conclusion, a high thrombotic risk among patients with aPL was indicated by the simultaneous presence of both LA and aCL, strongly positive aPL, and, among aCL, IgG-class antibodies.