Some prospective studies have shown that rabbit antithymocyte globulin was inferior to horse antithymocyte globulin as first-line therapy for patients with severe aplastic anemia. We retrospectively analyzed clinical outcome of 455 children with severe aplastic anemia who received horse antithymocyte globulin (n = 297) or rabbit antithymocyte globulin (n = 158) combined with cyclosporine as first-line therapy between 1992 and 2010. The response rates were comparable between horse and rabbit antithymocyte globulin groups at 3 months [46% (136/294) vs. 42% (66/153), p = 0.55] and 6 months [60% (178/292) vs. 55% (87/143), p = 1.0]. Using multivariate analysis, differences in antithymocyte globulin preparations were not associated with response rates. However, 2-year and 10-year overall survival rates in the horse antithymocyte globulin group were significantly better than in the rabbit antithymocyte globulin group (2-year overall survival: 96% vs. 87%, 10-year overall survival: 92% vs. 84%, p = 0.004). Using multivariate analysis, use of rabbit antithymocyte globulin was a significant adverse factor for overall survival (hazard ratio = 3.56, 95CI, 1.53-8.28, p = 0.003). Rabbit antithymocyte globulin caused more profound immunosuppression, which might be responsible for the higher incidence of severe infections. Considering that there are no studies showing superiority of rabbit antithymocyte globulin over horse antithymocyte globulin, horse antithymocyte globulin should be recommended as a first-line therapy. However, our results justify the use of rabbit antithymocyte globulin as first-line therapy if horse antithymocyte globulin is not available.