Movement between estuarine and marine habitats is a common occurrence among many exploited fish species, often associated with recruitment and/or spawning, but individual-level variation in migratory strategies can complicate the development of simple, tractable models of migration. Elemental profiles from otoliths are often used to reconstruct fish movements, given the well-established relationship between strontium, barium and salinity. However, interpretation of habitat-use and movement can be complicated by spatial and temporal variation in ambient elemental concentrations within- and among habitats and populations. Using otolith elemental profiles from tailor (Pomatomus saltatrix) along the east and west coast of Australia, we demonstrate how hidden Markov models can be used to estimate the age (year class) when transitions between estuarine and marine habitats occur, while accounting for (i) individual-level variation in movement strategies, and (ii) population-level variation in how estuarine and marine habitats are characterized. This application adds to the statistical toolkit for the reconstruction of fish movements using otolith elemental profiles by providing a framework to account for the natural sources of variation expected in these data (e.g., individuals, habitats, populations). Ultimately, we believe this will contribute to enhanced inferences regarding how exploited species move between estuarine and marine habitats.