Forty-one postmenopausal patients with abnormal uterine bleeding were scheduled for endometrial sampling in a week from the measurement of endometrial thickness by transvaginal ultrasonography. If the cut-off limit was 5 mm, the endometrial thickness was correlated with the presence of pathologic endometrium with a sensitivity of 87.5%, a specificity of 76%, a positive predictive value of 70% and a negative predictive value of 90.5%. Endometrial hyperplasia and polyp represented the two false negative cases. Waiting for a study on a sufficiently large series, the authors think that also an endometrial thickness of a few millimeters doesn't justify the non-recourse to endometrial biopsy in the presence of repeated haemorrhagic episodes or specific risk factors.