Tree swallows (Tachycineta bicolor) were captured and banded at six sites that differed in polychlorinated biphenyl (PCB) contamination levels in the Housatonic River watershed, western Massachusetts, USA, from 2000 through 2004 to test the prediction that apparent survival rates of females in more contaminated areas were lower than those from less contaminated areas. We also tested whether plumage coloration affected over-winter survival and whether concentrations of PCBs in eggs differed between birds that did and that did not return the following year. Apparent survival rates were calculated using mark—recapture methods and compared using Akaike's Information Criterion. Model-adjusted survival rates ranged from 0.365 to 0.467 for PCB-contaminated females and between 0.404 and 0.476 for reference females. Models with either survival or capture probability modeled as functions of treatment (degree of PCB contamination), year, and age received some support. The model-averaged parameter estimate reflecting a treatment effect for high-PCB birds was negative ( = −0.046, SE() = 0.0939). Fifty-four percent of the total model weights involved models in which survival was a function of PCB treatment. Eggs were collected for contaminant analyses from a random sample of females that did and that did not return the following year. Concentrations of total PCBs were the same or higher in the eggs of females that returned compared to the eggs of those that did not return at both the highly and the moderately contaminated PCB sites. This may have resulted from higher-quality females with higher lipid reserves being more likely than lower-quality females to return the following year. Percentage lipid was positively correlated with total PCBs in eggs. Survival rates were similar among swallows with brown versus blue plumage.