The development of a new, risk-averse harvest policy for the Northwestern Hawaiian Islands lobster trap fishery, based on computer simulations, is described. The analysis results from a 1994 expert panel review recommendation to develop a new harvesting strategy that incorporated uncertainty in the stock assessment and variability in population processes, with the dual goals of achieving industry objectives (high average catch and catch stability) and low risk of recruitment overfishing. Monte Carlo projections of an age-based simulation model which allows for systematic, process, random and measurement error, as well as autocorrelation in recruitment innovations, were used to compare harvesting strategies and to assess their effects relative to the risk of overfishing over a range of discard mortality and retention scenarios. The expected effects of alternative strategies and consequences of uncertainty were presented to the Western Pacific Regional Fishery Management Council for evaluation, and a constant harvest rate strategy, subject to a 10% overfishing risk, was chosen. Because of the presumed high rate of discard mortality in the fishery, which has since been verified, the Council also chose a retain-all retention policy. The constant harvest rate associated with these objectives is 13%. 1999 International Council for the Exploration of the Sea