Central nervous system (CNS) involvement in systemic lupus erythematosus (SLE) remains difficult to diagnose, particularly since structural abnormalities may not be revealed by magnetic resonance imaging (MRI). Glucose utilisation was measured by positron emission tomography (PET) in 35 SLE patients to detect signs of CNS involvement. The patients were examined by a standardised neurological examination, a battery of tests to evaluate neuropsychological performance and MRI. Antineuronal antibodies were determined to investigate their putative role in CNS involvement in SLE. Twenty patients had distinct neurological (17) and/or psychiatric (3) symptoms. Ten patients had pronounced cognitive impairment. Neurological and cognitive deficits were thus found to be unrelated disorders in SLE. Global glucose utilisation of SLE patients did not differ significantly from that of normal controls, nor were differences found between SLE patients with or without neurological or cognitive abnormalities. On MRI of the brain, the number and size of white matter lesions correlated with the presence of neurological deficits but were unrelated to the severity of cognitive impairment. Within the normal range, lower global glucose utilisation tended towards lower values with increasing number and size of white matter lesions. Patients with lesions larger than 8 mm also showed distinctly increased IgG anticardiolipin antibody titres, whereas measuring antineuronal antibodies did not reveal any relation to the variables investigated. We conclude that the demonstration of CNS lesions by MRI can contribute confirmatory evidence for CNS involvement in SLE, but PET or the presence of antineuronal antibodies adds little if any information beyond that obtained by clinical examination, neuropsychological testing, and MRI.