Flat oyster (Ostrea edulis L.) beds were once a dominant habitat type in the Dutch Delta and North Sea, but overharvesting resulted in soft sediment habitats devoid of oysters. Natural recovery of oyster beds will be slow if the natural substrate is lost and therefore, many oyster restoration projects have been set up worldwide. One way to enhance the success rate of restoring flat oyster beds is adding substrate at the moment that larvae are ready to settle. For this, more insight into the drivers of the timing of larval release is needed, which was the aim of this study. Generalized Additive Models (GAMs) were created based on historical data form the Oosterschelde and Lake Grevelingen of the abundance of swarming flat oyster larvae. It was shown that the temperature explains major part of the variation of larval occurrence. The addition of the temperature sum gave best results in the Oosterschelde. It was shown that the first peak in number of oyster larvae was predicted at a temperature sum of 576 degreedays. In Lake Grevelingen daily temperature yielded higher deviance explained values.Furthermore, the lunar cycle also contributed to the timing of larval release in the Oosterschelde, but not in the Lake Grevelingen, most likely since tides are absent in this waterbody. Chlorophyll-a partly explained larval occurrence in Lake Grevelingen, suggesting food abundance is another driving factor in the timing of gametogenetic processes of flat oysters. Furthermore, day-in-year and mean temperature also contribute significantly to the timing of larval swarming in both water bodies. When validated, this information can be used to predict theoptimal time window of deployment of substrate for spat settlement in order to increase the success rate of oyster bed restoration.