The purpose of this study was to compare the long-term effectiveness among danazol, corticosteroids, cytotoxics, and dapsone in the treatment of hematological manifestations of systemic lupus erythematosus (SLE). Medical charts of all patients seen at the Rheumatic Disease Unit from January to December of 1998 were reviewed. Patient characteristics, disease and treatment information were collected. The main outcome measures were the cause of and time to discontinuation of drugs used to treat hematological manifestations of SLE resulting from all causes, mainly toxicity and inefficacy or both. Bivariate analysis including one-way ANOVA and chi2 tests were used to compare differences between means and proportions, respectively. Survival curves among the different drugs were evaluated using the Kaplan-Meier method. Multivariate analysis (Cox-regression) was used to adjust for potential confounders. After all medical records were reviewed 41 cases were eligible. Two cases had hemolytic anemia, 34 had thrombocytopenia, and five had both. These cases had received a total of 121 cycles of treatment at different times and they represent the study population (corticosteroids n = 37, danazol n = 51, citotoxic drugs n = 29, and dapsone n = 4). Crude rates of discontinuations due to any cause, toxicity and inefficacy werenot statistically significant among the drugs. However, the Kaplan-Meier curves showed statistically significant difference for discontinuations due to all causes as well as inefficacy. Prednisone and cytotoxic drugs had the lowest probability of continuation. In contrast, there were not statistically significant differences among the drugs with respect to first relapse. This is the first study examining the long-term termination rates of several drugs used to treat hematological manifestations of SLE. Using rates of discontinuation adjusted for time there were statistically significant differences among the drugs. Danazol had the highest probability of continuation.