A hydrodynamic model with particle tracking was used to create individual-based simulations to describe larval fish dispersal through the restored Williamson River Delta and into Upper Klamath Lake, Oregon. The model was verified by converting particle ages to larval lengths and comparing these lengths to lengths of larvae in net catches. Correlations of simulated lengths with field data were moderate and suggested a species-specific difference in model performance. Particle trajectories through the delta were affected by wind speed and direction, lake elevation, and shoreline configuration. Once particles entered the lake, transport was a function of current speed and whether behavior enhanced transport (swimming aligned with currents) or countered transport through greater dispersal (faster random swimming). We tested sensitivity to swim speed (higher speeds led to greater dispersal and more retention), shoreline configuration (restoration increased retention relative to pre-restoration conditions), and lake elevation (retention was maximized at an intermediate elevation). The simulations also highlight additional biological questions, such as the extent to which spatially heterogeneous mortality or fish behavior and environmental cues could interact with wind-driven currents and contribute to patterns of dispersal.