The age composition of the catch and the growth curve of a stock are fundamentally important in fish stock assessment, but these estimates are subject to various sources of uncertainty. Using the Pacific bluefin tuna (Thunnus orientalis) fisheries in the waters off Taiwan as an example, we developed a Monte Carlo simulation model to evaluate the effects of four otolith sampling methods (random otolith sampling, ROS; fixed otolith sampling, FOS; proportional otolith sampling, POS; and reweighting otolith sampling, REW), and ageing error (bias and imprecision) on estimations of age composition and growth curves. The results indicated that FOS has the lowest sampling accuracy, POS performs the best and that ROS is a more efficient method with lower estimation error. For an imprecise reader, the centre (median) of multiple age reads is a useful method to obtain accurate and precise estimates. Ageing bias had greater effects on the estimation of age composition and growth parameters than ageing imprecision or the selection of otolith sampling methods. In most cases, 500 otoliths should be an adequate sample size and could be the guideline for the biological sampling program of the T. orientalis Catch Documentation Scheme.