Uterine smooth muscle tumors (SMTs) are common pelvic tumors in women, and most of them are diagnosed as usual leiomyoma (UL). Exclusion of malignant disease is important in the management of SMTs. However, differentiation of SMTs remains difficult. In this study, we aimed to improve the preoperative diagnosis of SMTs. We examined 21 ULs, 7 atypical leiomyomas (ALs), and 6 leiomyosarcomas (LMSs), all of which were diagnosed by uterine tumor biopsy. Immunohistochemical findings (low-molecular-mass polypeptide 2 (LMP2) and Ki-67) and clinical features (serum lactate dehydrogenase level and menopause) were evaluated. Statistically significant differences in the expression of LMP2 and Ki-67 were observed between UL and AL and between UL and LMS. The combined LMP2 and Ki-67 score was significantly different between UL and AL, between UL and LMS, and between AL and LMS. The combined immunohistochemistry and clinical findings score (total score) was also significantly different between pathological types. The findings of this study suggest that the accuracy of the preoperative diagnosis of SMTs may be improved by using a combination of immunohistochemical and clinical findings.