Fluctuations in the titers of anticardiolipin antibodies (aCL) have been reported in systemic lupus erythematosus (SLE) patients, but their relation with thrombosis is not completely understood. Prospective inception cohort of 237 patients with SLE (American College of Rheumatology criteria). Positivity for antiphospholipid antibodies (aPL) was defined according to Sapporo criteria. aCL was defined as persistently positive when more than two-thirds of the determinations were positive during follow-up. Patients were classified into four groups: A [positive lupus anticoagulant (LA)], B (negative LA and persistently positive aCL), C (negative LA and transiently positive aCL) and D (negative LA and aCL). Of these 237 patients, 211 (89%) patients were women. Median age at diagnosis and follow-up were 32 (2-78) and 10 (1-31) years, respectively; 33 (13.9%), 23 (9.7%), 42 (17.7%) and 139 (58.6%) patients were classified in groups A, B, C and D, respectively. Thirty (12.6%) and 23 (9.7%) patients suffered arterial and venous thrombotic events, respectively. Adjusted risk for arterial thrombosis was increased in groups A [odds ratio (OR) 15.69, 95% confidential interval (CI) 4.79-51.42, P < 0.001] and B (OR 7.63, 95% CI 2.00-29.08, P = 0.003), but not in group C when compared with group D. Adjusted risk of venous thrombosis was increased in group A (OR 4.24, 95% CI 1.36-13.20, P = 0.013), but not in groups B or C when compared with group D. Risk of thrombosis is not increased in SLE patients with negative LA and transiently positive aCL, even fulfilling Sapporo laboratory criteria, when compared with aPL-negative SLE patients.