The Panulirus ornatus stock in a 25000 km 2 area of Tortes Strait was estimated by making visual counts of the number of lobsters in strip transects. Pilot studies in 1988 to assess the feasibility of a full-scale survey and optimize the sampling design showed that: 4 x 500 m transects were the most cost-effective of the different sizes trialled; two transects per location comprised the most optimal allocation of replication; and 300 locations were necessary to achieve a 95% confidence interval of _+ 10 % of the mean density found in the pilot study. Satellite imagery was used to map habitats in Torres Strait, and areas likely to be inhabited by lobsters were classified broadly into three strata: windward reef slope, submerged reef, and deep areas. The 300 locations were allocated to each stratum in proportion to its area and the estimated variance of lobster abundance within it; once allocated, the locations were positioned at random within each stratum. The main survey was undertaken over a period of 7 wk in May-June 1989, and the resulting estimate of lobster abundance was ~ 14 million with a 95% confidence interval of +21%. The surveyed population was sampled concurrently to determine its size structure: the pre-fishery year-class comprised 43 % of the population; lobsters greater than legal-size comprised 57 % and their average tail weight was 346 g. Thus, the estimate of stock size for the study area was 2200 to 3350 t tail weight, which is roughly ten-fold greater than the annual catch of about 250 t. The current catch is approaching the lower estimates of potential yield, calculated using simple maximal sustainable yield estimators, which suggests that the fishery is unlikely to be under threat at present and may support greater effort.