Geophagy, the consumption of soil, is common in many species, but the drivers of geophagy are not well understood. The best-studied example of avian geophagy is the parrots of the western Amazon Basin, but even here, there is debate over what drives the behaviour. There are two possible explanations: (1) extra nutritional demands of reproduction drive an increase in geophagy, which would predict that geophagy should be highest during the breeding season, and (2) consumption of naturally toxic plant foods increases the need for the toxin protection effects of soil, which would predict that geophagy should be highest when food availability is low and animals are forced to consume more toxic foods. We used long-term data from lowland Amazonia to compare seasonal fluctuations in rainfall, food availability, parrot breeding and parrot geophagy, and conducted novel tests of these hypotheses. Our analyses of annual patterns suggested that seasonal changes in rainfall drive plant fruiting, the resulting food availability patterns drive the timing of parrot breeding, and breeding drives seasonal patterns of geophagy. Surprisingly, chicks of the largest psittacine species fledged as food supplies approach their annual lows, suggesting that future climatic changes that alter peaks in food availability could have unexpected impacts on the reproduction of large psittacines in this system. Our tests found no evidence to support the toxin-protection hypothesis. Instead, we found that the peak of geophagy occurred during species' breeding seasons, which strongly supports the supplemental nutrients hypothesis. Our findings join a growing body of biochemical, physiological, behavioural, ecological and biogeographical evidence suggesting that a need for sodium is driving soil consumption in this classic model system of avian geophagy.