In Manitoba, Warbling Vireos (Vireo gilvus) rejected eggs of the Brown-headed Cowbird (Molothrus ater) from 1 naturally parasitized nest and 16 experimentally parasitized nests studied in 1992 and 1993. Vireos rejected single, real cowbird eggs in 16 cases by ejection and 1 by desertion. Observations at nests during the first hour after parasitism showed that: (1) females carried out four ejections, whereas two of three unsuccessful ejection attempts were by males; and (2) the four observed egg removals were by puncture-ejection (entire cowbird egg removed or pieces of shell removed after egg contents consumed). These are the first observations of puncture-ejection by the Warbling Vireo, which is now the smallest species (15 g) known to eject cowbird eggs in this manner. Two competing hypotheses, evolutionary lag and evolutionary equilibrium, have been advanced to explain acceptance of brood parasitism by hosts parasitized with nonmimetic eggs. One prediction of the equilibrium hypothesis, that small hosts incur greater costs than large hosts when ejecting eggs, was examined by comparing the costs of rejection at experimentally parasitized nests of Warbling Vireos and larger Northern Orioles (Icterus galbula, 33 g). Warbling Vireos lost or damaged 5 of their own eggs while rejecting the 17 cowbird eggs, or 0.29 vireo eggs for every cowbird egg rejected. The cost to Northern Orioles of this behavior was 0.38 oriole eggs per cowbird egg ejected, which was not significantly different from the cost incurred by Warbling Vireos. My results by default support the evolutionary-lag hypothesis because the smaller species did not incur greater costs during puncture-ejection than the larger species. Warbling Vireos in and east of the Great Plains, V. g. gilvus, reject cowbird eggs, whereas individuals in a population west of the Great Plains, V. g. swainsonii (12 g), apparently accept the eggs. The different responses to cowbird eggs possibly reflect differences between two sibling species that have been in contact with cowbirds for different lengths of time, assuming that individuals in the western population are not below the minimum value for size of an ejecter.