Effective spatial management of marine species requires informed planning, as well as ongoing assessment. For mobile species such as fish, knowledge of the scale and variation in movement is central to key planning decisions, such as the size and shape of marine reserves and the interpretation of the response of protected populations. For example, populations of species that require large areas of habitat may not show increases in abundance inside small reserves, but calculating optimal reserve size is complicated by individual variations in behaviour. Fish movements can be used to quantitatively inform marine reserve planning and assessment. An individual based numerical simulation model including acoustic telemetry and census data was used to simulate changes in populations of snapper Pagrus auratus in north-eastern New Zealand. Four behavioural categories and offshore migration were used to represent the observed variability in movement. Age-structures of modelled fish populations in fully exploited areas, marine reserves and virgin populations differed substantially. However, the population structure within reserves resembled a fully fished population more closely than an unfished population. Due to the range of movement types shown by snapper, fish were not ‘locked up’ by reserves, and fish with centres of activity based in reserves were predicted to have a relatively high chance of being caught outside these reserves. Furthermore, the model showed that the response of fish populations within marine reserves was dependent on levels of exploitation in fished areas. For snapper in coastal reef areas, reserves c. 40 km2 or more may be required to achieve abundances > 50% of the unfished stock. On balance, while marine reserves with sizes similar to Leigh and Tawharanui (c. 5 km2) can achieve significant levels of protection for snapper, they are too small to fully protect resident reserve snapper populations.