We performed echocardiography prospectively 4.9 +/- 0.7 years apart (mean +/- SD), in 74 patients with systemic lupus erythematosus. On the basis of the first study, the patients were distributed in four groups according to the type of valvular involvement: 7 patients had vegetations (Libman-Sacks endocarditis; group 1); 6 patients had rigid and thickened valves with stenosis, regurgitation, or both (group 2); 5 patients had miscellaneous forms of valvular involvement without valvular dysfunction (group 3), as did the 60 controls; and 56 patients had no valvular disease (group 4). The overall prevalence of clinically important valvular disease (groups 1 and 2) was 18 percent. Patients in group 1 were younger than those in group 2 (33.5 +/- 16.7 vs. 47.8 +/- 17.6 years; P less than 0.05), had a shorter mean duration of lupus (4.8 +/- 2.2 vs. 10.7 +/- 6.4 years; P less than 0.001), and had received a smaller cumulative dose of steroids (21.5 +/- 13.1 vs. 79.5 +/- 63.4 g of methylprednisolone or its equivalent; P less than 0.05). During the five-year follow-up, one patient in group 1 and five in group 2 required valve surgery, no patient in group 3 had valvular dysfunction, and five patients in group 4 had mild valvular lesions. We conclude that clinically important valvular involvement in systemic lupus is relatively frequent and sometimes requires surgery. Echocardiography can identify a subset of lesions (valvular thickening and dysfunction), other than verrucous (Libman-Sacks) endocarditis, that are prone to hemodynamic deterioration.