OBJECTIVE: verify the predictive value of transfontanellar ultrasonic exams in premature babies on determination of altered neural psychomotor development. METHOD: prospective cohort.: Ninety-nine premature infants with birth weights of 1800 g or less and gestational ages of less than 37 weeks who had had neonatal transfontanellar ultrasound scans when they were in intensive care were studied. The Denver II test was used to assess neurodevelopment. The chi-square test and logistic regression were used in the bivariate and multivariate analyses, respectively, and revealed associations between the socioeconomic characteristics and dependent variable. Lastly, an analysis was carried out to determine positive and negative predictive values. RESULTS: the population consisted of children with a corrected age of 12 months, average birth weight of 1320 g and average gestational age of 31 weeks and 3 days. Ultrasound abnormalities were detected in 49.4% of the children. Of the tests carried out, 34.3% yielded unfavorable results. Among the risk factors for delayed development, abnormal ultrasound results and family income were statistically significant. The positive and negative predictive values of the transfontanellar ultrasound scans for altered neurodevelopment were 51,02 % and 82 %, respectively. When the variable "family income" was used together with "altered ultrasound results", the positive predictive value increased to 90% and the negative predictive value fell to 71.91%. CONCLUSION: inclusion of "family income" is an effective way of increasing the predictive value of abnormal transfontanellar ultrasound results for altered neurodevelopment in premature infants.