The Atlantic blue crab Callinectes sapidus is an important predator in estuarine ecosystems and supports a large commercial fishery throughout its range. Settlement of blue crab megalopae in juvenile nurseries is episodic and is controlled by the combined effect of buoyancy- and wind-driven transport processes. Modeling studies have succeeded in hind-casting the temporal pattern of megalopal settlement but not the magnitude of individual settlement events. This shortcoming has been attributed to the lack of information concerning larval release by the adult spawning population. In this paper we present results of a set of field and modeling studies designed to test the hypothesis that temporal variation in release of larvae controls the size of settlement events. Field work consisted of collections of ovigerous female blue crabs during their spawning migration in Delaware Bay (∼38.5N; 75.1W). Egg samples were taken from each female, and the predicted date of hatching was determined using a morphological index developed as part of the study. Spectral analysis of the time series of daily larval-release events showed maximum density at low-frequency (period ≈ 40-70 days) and reflected the expected seasonality in spawning activity. There was only modest variance at higher frequencies (period ≈ 3–5 days), and the low-frequency harmonic was responsible for ∼33% of the overall variability. A series of modeling experiments was performed for the years 2002-2005. These experiments compared the observed larval release pattern in 2005, a constant release pattern, and a number of synthetic time series that incorporated the measured high- and low-frequency variations obtained from the observations. Results of these experiments showed that temporal variation in larval release can have measurable effects on the magnitude of individual settlement events. The effects on total annual settlement of high- and low-frequency variations are comparable (5–26% coefficient of variation) for any given year; however, these effects are relatively small when compared to physically-driven interannual variation in simulated settlement (81% coefficient of variation).