Despite recent advances in technology, it remains difficult to connect breeding and non-breeding areas of populations of migratory organisms due to the challenges of year-round tracking. Here, we used the Eurasian reed warbler Acrocephalus scirpaceus, a passerine with a pronounced migratory divide to demonstrate the promise of integrating several sources of information within the Bayesian modelling framework for the study of migratory connectivity. To this end, we combined data from stable hydrogen isotope ratios (δ2H) of feathers, ring recoveries, and the geographic delineation of sub-populations on either side of the migratory divide. Feather δ2H measurements from local juvenile birds sampled across the breeding range tightly correlated with amount-weighted mean annual precipitation δ2H values predicted for the natal sites. Predicted natal origins of birds intercepted en route in the Mediterranean region largely differed among the five stopover sites. Thanks to the different migratory pathways used by different breeding populations and the existence of a migratory divide, we were able to effectively narrow the assigned regions of origin. Our results show that spatial resolution of likelihood-based assignments of geographic origins based on δ2H measurements may improve significantly when prior probabilities derived from population-specific migratory directions are included. Integrating information from stable isotopes, ring recoveries, geolocators and other sources within the Bayesian modelling framework will provide an extremely useful toolbox for the study of animal movements in the future.