Larval dispersal models increasingly include larval behavior rather than assume passive larvae. However, few include a range of dispersal-relevant behaviors. Fewer still are tested. Many models over-generalize adult behaviors such as seasonal, lunar and diel timing or location of propagules release. Even if empirically-based behavior is included, variation among individual propagules is seldom included: rather, mean behavior is used, unrealistically constraining dispersal outcomes. Our model for Great Barrier Reef (GBR) fishes uses empirical measures of larva vertical distribution, swimming and orientation, and variation in and ontogeny of them. Only virtual larvae with sufficient swimming ability to overcome currents within ‘detection zones’ around reefs can settle. Propagules are released at empirically-determined times and places. Egg buoyancy is included. Hydrodynamics are a 3D development of the 2D James, et al. (2002) model. We tested three model scenario predictions (1- all empirical larval behavior; 2 - hypothetical SE swimming; and 3 - passive larvae) for a grouper (Plectropomus maculatus) in the southern GBR using genetic parentage techniques that matched recruits at the end of the 2011/12 spawning season with adults from spawning aggregations (Williamson et al. in review). Observed dispersal direction and distance best fit model Scenario 1 predictions.