OBJECTIVE: To evaluate the relative prognostic value of ultrasound findings during the first-trimester scan by univariate and logistic regression analysis in a group of asymptomatic women. STUDY DESIGN. A total of 125 asymptomatic pregnant women with a singleton conceptus, concordance between menstrual age and crown-rump length, and documented fetal activity, 25 for each gestational age between 6 and 10 weeks, were enrolled in the study and underwent a transvaginal sonographic examination. The following data were collected: yolk sac mean diameter and volume, gestational sac mean diameter and volume, fetal heart rate, maternal age, gestational age and presence of a retrochorial hematoma. The outcome variable was abortion, defined as pregnancy loss at any time up to 20 weeks' gestation. Normograms were constructed for volumes, mean diameters and fetal heart rate. Receiver operator characteristic curves were performed in order to dichotomize maternal and gestational age. Univariate analysis was performed by Fisher's exact test. Logistic regression was performed to test the relationship between independent variables and pregnancy outcome. RESULTS: In univariate analysis, the variables significantly associated with spontaneous abortion were maternal age > 34 years, yolk sac volume outside the 5th to 95th percentile, gestational sac volume < the 5th percentile and fetal heart rate outside the 5th to 95th percentile. In regression analysis only maternal age > 34 years, gestational sac mean diameter < 5th percentile and fetal heart rate outside the 5th to 95th percentile were significant in predicting abortion. CONCLUSION: Our data suggest that new three-dimensional parameters are of no clinical benefit in the prediction of abortion in nonbleeding, first-trimester pregnancy, when conventional sonographic parameters are used. (J Reprod Med 2003;48:252-256)