This article provides further insights into the properties of momentum trading strategies using information from the Australian market. Based on a methodology that avoids the look-ahead bias of many momentum studies that employ monthly data, we confirm the existence of a momentum effect in Australia. In contrast to previously reported results, momentum is stronger amongst larger firms in the Australian market and buying 'winners' generates higher returns than shorting 'losers'. We find strong seasonal influences which are consistent with the tax selling hypothesis and institutional 'window dressing'. In addition, we show that momentum returns are highly variable over time. Specifically, the momentum strategies employed in the late 1990s generate higher returns than those in the early 1990s. Some aspects of the effect are quite different from those previously observed in other markets and this is useful for testing theories about the causes of momentum out of sample. We use information on the intertemporal performance of 'winners' and 'losers' in different market states to determine which of a number of behavioural theories are most predictive of the observed movements of the Australian market. The evidence indicates that models based on the disposition effect better fit the observed data than models based on an overreaction bias.