Miniaturized light-level geolocators may revolutionise the study of avian migration. However, there are increasing concerns that they might negatively affect fitness. We investigated the impact of two miniaturized geolocator models (SOI-GDL2.10, deployed in 2010, and SOI-GDL2.11, deployed in 2011) on fitness traits of the barn swallow Hirundo rustica, one of the smallest migratory species to which geolocators have been applied to date. The 2011 model was lighter (by 0.09 g) and had a shorter light stalk compared to the 2010 model. Using data from 640 geolocator and 399 control individuals from three geographical populations, we found that geolocators reduced annual survival probabilities (control birds: 0.19–0.63; geolocator birds: 0.08–0.40, depending on year, sex, and how birds that lost the device were considered), with more markedly negative effects on females equipped with the 2010 model. In addition, among birds equipped with the 2010 model, onset of reproduction in the subsequent year was delayed (by 12 d) and females laid smaller first clutches (by 1.5 eggs, i.e. a 30% reduction) compared to controls. Equipping parents with geolocators while they were attending their brood did not affect nestling body mass or fledging success. A reduction of geolocator weight and drag by shortening the light stalk slightly enhanced the survival of females but not that of males, and mitigated the negative carry-over effects on subsequent reproduction. Our study shows that geolocators can have a negative impact on survival and reproduction, and that even minor differences in weight and drag can make the difference. We suggest that studies aiming at deploying geolocators or other year-round tagging devices should be preceded by pilot experiments to test for fitness effects.