Thirty-five patients with systemic lupus erythematosus (SLE) have been monitored clinically and serologically for up to three and a half years. The data collected was analysed by computer using the Statistical Package for the Social Sciences Program. The patients were categorized into severely active, moderately active and inactive disease groups and associations between clinical state and laboratory tests were sought. No single test was found to distinguish, reliably, the clinical groups but levels of circulating immune complexes (by polyethylene glycol precipitation), platelet count and ESR were shown to distinguish inactive from active disease (p less than 0.05). Severely active disease was distinguished from the less active forms by circulating immune complex levels (Clq solid phase assay) double-stranded DNA binding, lymphocyte count, and CH50 estimations (p less than 0.05). Tests were found to differ considerably in their ability to reflect disease activity when patients were separated into sub-groups according to the major clinical features (eg. arthralgia, renal disease and vasculitic rash). However, those patients with cerebral manifestations and thrombocytopenia proved to be the most difficult to assess. Discriminant function analysis showed that a maximum of 44 per cent of cases could be correctly classified into their clinical grades when combinations of four out of five laboratory tests were used. These results emphasise the continuing need for better tests to monitor the course of SLE.