Seventy unselected patients with systemic lupus erythematosus (SLE) were studied to determine the prevalence of cognitive impairment and the association with other clinical variables. Twenty-five patients with rheumatoid arthritis (RA) and 23 healthy subjects were used as controls. All patients were evaluated with a battery of standardized neuropsychological tests to determine ability in 8 areas of cognitive function. Clinically overt neuropsychiatric (NP) SLE, cumulative disease manifestations and concurrent medications were documented. In patients with SLE, generalized disease activity was expressed using the SLE disease activity index. Cognitive impairment was identified in 15/70 (21%) patients with SLE, 1/25 (4%) patients with RA and in 1/23 (4%) healthy subjects (p = 0.042). The prevalence was higher in patients with active NP-SLE at the time of assessment (2/5, 40%) compared to patients with inactive NP-SLE (2/10, 20%) but was also increased in those patients who had never had known clinical NP-SLE (11/55, 20%). A history of serositis (p = 0.015), active SLE (p = 0.064) and corticosteroid use (p = 0.027) at the time of assessment were more common in patients with cognitive impairment. The results suggest that cognitive impairment is increased in patients with SLE. It may occur independently of clinically overt NP-SLE and is more common in patients with active disease who are receiving corticosteroids.