The dose of trimethoprim-sulfamethoxazole (TMP-SMX) for the treatment of Pneumocystis pneumonia (PCP) in patients without human immunodeficiency virus (HIV) infection has not been verified. The aim of this study was to investigate the efficacy and toxicity of a low-dose TMP-SMX regimen in such patients. A retrospective study was conducted in four hospitals. We reviewed the medical records of patients with PCP but not HIV (non-HIV-PCP) who were treated with TMP-SMX between 2003 and 2016. The patients were divided into conventional-dose (TMP, 15 to 20 mg/kg/day) and low-dose (TMP, <15 mg/kg/day) groups after patients who received high-dose (TMP, >20 mg/kg/day) treatment were excluded. Grouping was done according to a correction dose, which was based on renal function. Eighty-two patients had non-HIV-PCP. The numbers of patients who received high-, conventional-, and low-dose treatments were 5, 36, and 41, respectively. Kaplan-Meier analysis for death associated with PCP showed no statistically significant difference in survival rates between the conventional- and low-dose groups. Ninety-day cause-specific mortality rates were 25.0% and 19.5% in the conventional-dose and low-dose groups ( P = 0.76), respectively. Adverse events that were graded as ≥3 according to the Common Terminology Criteria for Adverse Events (version 4.0) (National Cancer Institute, 2010) were 41.7% and 17.1% in the conventional-dose and low-dose groups ( P = 0.02), respectively. Moreover, vomiting ( P = 0.03) and a decrease in platelet count ( P = 0.03) occurred more frequently in the conventional-dose group. Treatment of non-HIV-PCP with low-dose or conventional-dose TMP-SMX produces comparable survival rates; however, the low-dose regimen is better tolerated and associated with fewer adverse effects.