Animals are often faced with complex movement decisions, particularly those that involve long-distance dispersal. Partial migrations, ubiquitous among all groups of vertebrates, are a form of long-distance movement that occurs when only some of the animals in a population migrate. The decision to migrate or to be a resident can be dependent on many factors, but these factors are rarely quantified in fishes, particularly top predators, even though partial migrations may have important implications for ecosystem dynamics and conservation. We utilized passive acoustic telemetry, with a Brownian bridge movement model and generalized additive mixed models, to explore the factors regulating partial migration in a large marine predator, the tiger shark, throughout the Hawaiian Islands. Although sharks tended to utilize a particular "core" island, they also demonstrated inter-island movements, particularly mature females that would swim from the northwestern Hawaiian Islands to the main Hawaiian Islands (MHI). Immigration to another island was a function of season, sea surface temperature (SST), and chlorophyll a concentration. Our results predict that 25% of mature females moved from remote French Frigate Shoals atoll to the MHI during late summer/early fall, potentially to give birth. Females with core home ranges within the MHI showed limited movements to the NWHI, and immigration to an island was better explained by SST and chlorophyll a concentration, suggesting a foraging function. Dispersal patterns in tiger sharks are complex but can be considered a mix of skipped-breeding partial migration by mature females and individual-based inter-island movements potentially linked to foraging. Therefore, sharks appear to use a conditional strategy based on fixed intrinsic and flexible extrinsic states. The application of Brownian bridge movement models to electronic presence/absence data provides a new technique for assessing the influence of habitat and environmental conditions on patterns of movement for fish populations.