Bivalve larvae in mixed samples collected from the field have traditionally been identified through morphological differences among species. It is difficult, however, to use this method accurately because of overlapping size ranges and similar shapes of the larvae of many species. We used the molecular technique developed by Hare et al. (2000) to test the accuracy of morphological identification of Mercenaria mercenaria (L.) larvae from plankton samples collected from the Great South Bay and Coecles Harbor on Long Island, New York. We found that morphology is unreliable as the only means of identification for bivalve larvae in a mixed field sample, and a very high false positive rate of identification of M. mercenaria (100% of 71 larvae were misidentified). Morphological characteristics may be used to eliminate larvae from a field plankton sample, as the false negative rate for M. mercenaria was only 1.4% (n = 140). To determine larval bivalve densities accurately, other techniques in addition to those based on morphological characteristics, such as the molecular technique used in this study, must be used.