Aerial insectivorous birds have suffered steep population declines in North America over the last 60 years. A lack of information on migratory connectivity between breeding and non-breeding grounds for these species limits our ability to interpret factors affecting their population-specific trends. We determined likely Latin American non-breeding regions of Bank (Riparia riparia), Barn (Hirundo rustica) and Cliff (Petrochelidon pyrrhonota) swallow from populations across their breeding ranges. We used predicted feather hydrogen (δ2Hf) and carbon (δ13Cf) isoscapes for winter-grown feathers to indicate areas of highest probability of moult origin and incorporated these results into a cluster analysis to determine likely broad non-breeding regions. We also assessed variation in wing length among populations to determine the potential for this metric to differentiate population moult origins. We then investigated patterns of multi-isotopic (δ2Hf, δ13Cf, δ15Nf) and wing-length niche occupancy by quantifying niche size and overlap among populations under the assumption that broad niches were consistent with low within-species migratory connectivity and narrow and non-overlapping niches with higher connectivity. Multivariate assignment identified different non-breeding regions and potential clusters of moult origin generally corresponding to Central America and northern South America, eastern and south-central South America, and the western and southern part of that continent, with variation within and among populations and species. Separate niche space indicated different wintering habitat or areas used by species or populations whereas niche overlap indicated only potential spatial similarity. Wing length varied significantly among populations by species, being longer in the west and north for Bank and Cliff Swallow and longer in eastern Canadian Barn Swallow populations. Barn Swallow occupied consistently larger isotopic and wing length niche space than the other species. Comparisons among populations across species showed variable isotopic and wing-length niche overlap generally being greater within breeding regions and lower between western and eastern breeding populations supporting a general North American continental divide for all species with generally low migratory connectivity for all species. We present a novel approach to assessing connectivity using inexpensive and broad isotopic approaches that provides the basis for hypothesis testing using more spatially explicit expensive techniques.