We investigated movements of a western population of Veeries (Catharus fuscescens) breeding in the Okanagan region of British Columbia, Canada, in 2013–2014 using light-level geolocators. We tracked 9 individuals and incorporated a state-space Kalman filter model approach to estimate movement parameters. During migration, Veeries traversed the Rocky Mountains, Great Plains, Gulf of Mexico, and Caribbean Sea with stopovers generally closer to the shorter orthodromic (great circle) route than a loxodromic (straight line) route between breeding and first wintering grounds, particularly on fall migration. Birds initially settled in the south-central portion of the Amazon basin in Brazil at sites that were 666 ± 299 km apart, suggesting low migratory connectivity. Intra-tropical movements were observed for 8 of 9 (88.9%) birds, with second wintering sites an average of 1,447 ± 472 km to the northwest (initial bearing x̄ = 316 ± 16°). Veeries typically followed a pattern of loop migration at the Gulf of Mexico, with more birds using the Yucatan Peninsula to stop and reorient toward destinations on spring migration (n = 7) vs. fall migration (n = 2). Western Veeries follow a presumed ancestral (eastern) migration route, but this route is also the shortest (great circle) route between breeding and wintering grounds, even though this route was only ~100 km shorter than the straight line route. Eight Veeries (88.9%) underwent a post-breeding, pre-migratory movement up to 628 km (x̄ = 263 ± 152 km) away from breeding territories, possibly to molt. We encourage researchers utilizing light-level geolocators to apply similar state-space modeling approaches to reduce the influence of observers and erroneous location estimates on analysis and interpretation of geolocator data.