OBJECTIVE In this study we compare transvaginal sonography with MR imaging for use in detecting the depth of myometrial involvement by endometrial carcinoma. METHODS Forty-two consecutive patients with stage I endometrial carcinoma had transvaginal sonography and MR imaging at 0.5 T. All the patients had a hysterectomy within 1-10 days after the imaging studies. The results of histologic examination of the surgical specimen were considered the gold standard of the study. We compared transvaginal sonography and MR imaging for use in assessing myometrial invasion by endometrial carcinoma by means of the staging classification of the International Federation of Gynecology and Obstetrics: stage Ia (tumor limited to endometrium), stage Ib (invasion of less than half the myometrium), stage Ic (invasion of more than half the myometrium). The overdiagnoses and the underdiagnoses for both techniques were calculated. We also evaluated the sensitivity and specificity of the two techniques for assessing the presence of myometrial invasion (stage Ib + stage Ic) and the presence of deep myometrial invasion (stage Ic). The diagnostic indexes evaluated and the differences between them were analyzed by using McNemar's test and 95% confidence intervals. The staging diagnoses based on MR imaging and sonographic findings were compared with staging diagnoses based on histologic examination, and a score was assigned to each diagnosis: these scores were then evaluated with Wilcoxon's signed rank test for paired data. RESULTS Histologic examination showed that six of the 42 patients had tumor confined to the endometrium (stage Ia), 14 had involvement of the inner half of the myometrium (stage Ib), and 22 had involvement of the outer half of the myometrium (stage Ic). The staging was concordant between the two imaging techniques in 32 cases (concordance, 76%). Among the 10 discordant cases, diagnosis was correct in six cases for MR and four cases for sonography. Overall staging based on sonography was correct with respect to histologic staging in 29 cases (69%; 95% confidence interval, 52-81%). Five tumors (12%) were underdiagnosed and eight (19%) were overdiagnosed. Staging based on MR findings was correct with respect to histologic staging in 31 cases (74%; 95% confidence interval, 58-85%). Five tumors (12%) were underdiagnosed, and six (14%) were overdiagnosed. CONCLUSIONS In our experience, there is no difference in the staging diagnoses of transvaginal sonography and MR imaging. Also, concordance with histologic staging diagnoses and sensitivity and specificity indexes did not show statistical differences between the two techniques, although these last results have to be considered with caution because of the low power of the statistical tests.