Transvaginal ultrasound and serum concentration of human chorionic gonadotropin are the main methods of prognostic evaluation of early pregnancies. Although these pregnancy investigation methods have proven their benefits, the interpretation is fundamental for a correct diagnosis, in order to avoid interventions that may affect pregnancies that would normally have had a physiologic course, as well as to apply an appropriate management in the case of nonviable pregnancies. Recent studies have shown the need to adopt rigorous criteria that minimize false positive results and thus decrease the frequency of diagnostic errors. Over time, a series of ultrasound features have been described which have proven effective in predicting nonviable pregnancies in the first trimester (cranio-caudal length in relation to the presence or absence of cardiac activity, yolk sac diameter, gestational sac diameter in relation to presence or absence of embryo with cardiac activity, embryonic heart rate, subchorionic hematoma etc.). In the context of the new prospects for the development of artificial intelligence softwares, the present work focuses on the description of the precise ultrasonographic aspects aboded by the latest studies that propose certain cut-off values dependent on the gestational age of the ultrasound parameters in the first trimester, increasing the sensitivity and the specificity of ultrasound in the diagnosis of nonviable early pregnancies, thus aiming to improve the reproductive prognosis of couples with recurrent abortions. [ABSTRACT FROM AUTHOR]