We used DNA markers to determine the sexes of 449 breeding adult Black Terns (Chlidonias niger) captured and measured in southeastern Wisconsin from 2000 to 2005. Various morphometric discriminant functions were then generated to classify the sexes of these birds. Other objectives were to examine the effect of sample size on the predictive ability of discriminant function analysis (DFA), to evaluate the temporal stability in the accuracy of DFA for a specific population, and to test the geographic specificity of DFA by comparing separate breeding colonies in southeastern Wisconsin and also by applying a discriminant function, derived from an Oregon population, to terns in Wisconsin. The best discriminant function included head-plus-bill length and body mass, and correctly classified 81% of the birds in the complete sample. Within-pair comparisons increased the classification accuracy to 87%, because males usually were larger than their mates with respect to head-plus-bill length. Sample-size analysis indicated that a satisfactory discriminant function for Black Terns could be generated from a sample of only 10% of the population. Region-wide, classification accuracy was similar among years, indicating temporal stability, but the accuracy of prediction for females tended to erode over time at specific colonies. Discriminant functions generated from birds captured at one site performed equally well on neighboring populations, but the Oregon function performed poorly on the Wisconsin population. Our findings suggest that a discriminant function derived from morphometrics of a single population can be applied to other populations in a limited geographic region but its utility may decrease over time. Our results also validate the previous cautions that geographic variation in the biometry of a species precludes the use of a single discriminant function on widely disjunct populations.