The association between time devoted to homework and children's academic achievement has long been an issue of great debate. A small number of mainly correlational studies have been conducted into this issue in a primary school setting, but have produced somewhat mixed results. In this paper we contribute to this literature by investigating the relationship between time spent upon homework and children's outcomes across 24 countries. By using a student fixed-effects approach, capturing differences in homework time amongst the same student across different school subjects, we argue that our results are likely to be subject to less confounding than much of the existing literature. We find little evidence that the amount of homework time primary school children are assigned is related to their academic achievement. This holds true across a large number of countries, survives various robustness tests and does not vary by gender or socio-economic status. We interpret this finding as suggesting that the homework assigned to primary school pupils may not be adequate to produce a positive association, and needs be improved if this time-consuming activity is ever going to bring benefits for children's academic achievement.