In a spatiotemporally variable environment, plants use seed dispersal and dormancy to reduce risk. Intuition suggests that dormancy should be able to substitute for dispersal, so that dormancy will reduce the optimal mean dispersal distance, and previous theoretical studies using temporally uncorrelated environments have found this to be true. I show that in the presence of positive temporal correlations, dormancy instead increases dispersal: dormancy and dispersal are not interchangeable risk reduction mechanisms. Dispersal has both costs (seeds landing in unfavourable habitat) and benefits (seeds being in place to exploit newly favourable habitat). I discuss how the costs and benefits balance to determine optimal dispersal and how dormancy shifts this balance, causing dispersal to increase. I also find that an interaction between spatial and temporal correlations determines whether an evolutionarily stable dispersal distance exists at all and confirm the expectation that increasing the scale of spatial correlations causes dispersal to increase.