Light traps were used to sample small fish and squid from open waters in the central Great Barrier Reef lagoon. A total of 7203 fish, representing some 38 families, and 706 lologinid squid were caught during sampling periods October to January, 1988 to 1990. The fish catch was dominated by the family Pomacentridae (63 % of fish collected), with lower numbers of lethrinids (6.7 %), clupeids (6.3 %), mullids (3.8 %), and scombrids (2.7 %). Size-frequencies of the fish collected indicated that the light traps sampled late-stage larvae and pelagic juveniles exclusively. No effect of time of night on catch rate was detected. Light traps that were allowed to drift with prevailing water currents caught more fish than anchored traps; this unexpected result may be a function of the effect of current velocity on trap efficiency. Analysis of standard errorkample size curves suggested that optimum replication was achieved with 5 to 6 traps, but that reasonable precision could be obtained with 2 to 3 traps. Coefficients of variation among replicate traps were taxon-specific, ranging from 0.9 (for clupeids) to 0.2-0.1 (for pomacentrids). These values compare favourably with those obtained from trawl nets. Light traps have considerable potential for sampling nekton that are capable of avoiding conventional towed nets.