We sought to test our clinical impression that using a low dose methylprednisolone pulse (MEP; < or = 1500 mg over 3 days) in treating flares of systemic lupus erythematosus (SLE) was effective and associated with fewer serious infections. We retrospectively studied SLE patients who received MEP between 1989 and 2000. A 'low dose' group of 26 patients who had received 1-1.5 g and a 'high dose' group of 29 patients who received 3-5 g of MEP were identified. SLEDAI scores and prednisolone doses were recorded at the time of MEP pulses and 6 months later. All serious infections (requiring admission and i.v. antibiotics) occurring during this 6 month period and their outcomes were recorded. Both groups had similar demographic data, initial SLEDAI scores, i.v. cyclophosphamide use, and SLE organ involvement. Despite high- and low-dose MEP being efficacious in controlling disease activity (lowering of SLEDAI scores and subsequent prednisolone dose) there were only nine episodes of serious infection in seven patients in the low-dose group compared with 20 episodes in 17 patients from the high-dose group (P = 0.04). In both groups a majority of infections (75 and 77% in the high- and low-dose groups) occurred in the first month after MEP. Those with a low serum albumin (< 20 g/l) had an increased risk of mortality (OR 44, 90% CI 6.19-312.98) and a trend towards greater numbers of infections. Low-dose MEP was effective in controlling SLE flares and associated with fewer serious infections than traditional high-dose MEP.