Many data poor stocks, such as those of most deep-water fish species in the Indo-Pacific region, are characterized by limited (or absent) catch time series and mixed-species, fishery-dependent CPUE indices that are unlikely to constitute reliable indices of stock abundance. For these situations, several data-limited assessment approaches employing only information on species biology combined with size and/or age composition data, are available for producing estimates of stock status to inform fisheries management. Some of these approaches include mean length mortality estimators, catch curve and per recruit analyses, the length-based spawning potential ratio method and length-based Bayesian biomass method. All of these methods make relatively strong assumptions which, when violated, can impact on the reliability of assessment results. This presentation explores the implications, for data-limited assessments, of using alternative model assumptions relating to stock-recruitment dynamics, post-release mortality and (isometric vs hyperallometric) reproductive scaling. The utility of a form of equilibrium analysis for improving the biological realism of several assumptions in data-limited assessments is explored using simulations from a new R package.