The relative validities of three clinical assessment methods for estimating gestational age in newborn low birth weight infants were evaluated with reference to estimates based on the date of the mother's last menstrual period. For 1246 infants in eight diverse institutions, estimates based on physical criteria correlated more strongly with dates estimates, yielded estimates more similar on average to dates estimates, and yielded higher proportions of correct classifications of prematurity and small for gestational age than did estimates based on neurologic criteria or neurologic and physical criteria combined. These results support the use of physical criteria rather than neurologic or combined criteria for the clinical assessment of gestational age in low birth weight infants. In a subsample of 511 black and white infants, there were no ethnic differences in mean error of estimate for any of the three methods.