To determine the utility of magnetic resonance imaging (MRI) of the brain in diagnosing active neuropsychiatric disease in systemic lupus erythematosus (NP-SLE), a prospective study of 51 hospitalized systemic lupus erythematosus (SLE) patients during 64 separate episodes of suspected NP-SLE was initiated. In addition to standard hematology, chemistry, and serological tests, the workup included MRI in all patients. A computed tomographic scan of the brain was obtained in patients enrolled in the first year of the study. Of the 64 neuropsychiatric episodes, 42 were attributable to NP-SLE and 22 were attributed to causes other than SLE. Neuropsychiatric complaints unrelated to lupus included depression (n = 6), seizures (n = 5), headache (n = 3), altered mental status (n = 2), aseptic meningitis (n = 2), cardiovascular accident (n = 2), transient ischemic attack (n = 1), and vertigo (n = 1). The MRI was abnormal in 34 of 64 (53%) episodes. MRI abnormalities were more common in patients with focal neurological deficits (19/26) than in those without focal findings (15/38; P = .008) and in patients with nephritis (19/24) than in those without renal disease (15/40; P = .002). MRI abnormalities were as frequent in NP-SLE (25/42) as in cases with non-NP-SLE-related causes (9/22). Periventricular increased signal (PIS) was a frequent MRI finding (10/64). Enlargement of the prepontine cistern, an MRI finding not previously described in NP-SLE, was seen (14/64). Both findings were associated with the presence of hypertension and lupus nephritis. PIS similar to that seen in our patients has been described in otherwise healthy elderly individuals with risk factors for stroke, suggesting that vascular abnormalities may be important in the etiology of these lesions. In conclusion, abnormalities in brain MRI occur frequently in NP-SLE, especially in patients with focal neurological deficits. However, the presence of similar MRI abnormalities in SLE patients with neuropsychiatric symptoms and findings with non-SLE-related causes limits the specificity of the MRI for diagnosing NP-SLE.