Researchers commonly collect blood samples from wild birds, and most workers assume that blood sampling has no adverse effect on the birds' survival. Few studies, however, have done controlled comparisons among bled and non-bled individuals and estimated survival using modern statistical methodology. We used a data set on Cliff Swallows (Petrochelidon pyrrhonota) that included 2,945 bled and 7,822 non-bled birds captured at the same times and sites in southwestern Nebraska from 1986 to 2006 to estimate annual survival and recapture probabilities of each group. Blood was collected with brachial venipuncture in amounts varying from 0.3% to 1.2% of the birds' body mass. Apparent survival of bled birds was lower than that of non-bled birds: bled birds experienced a 21–33% reduction in average survivorship, depending on amount of blood taken and whether the individuals were resident at a fumigated (parasite-free) or non-fumigated colony at the time of sampling. The percent reduction in annual survival was higher for individuals at non-fumigated colonies. All effects of blood sampling applied only in the year after sampling, and there were no effects in later years. Our results suggest that brachial blood sampling is not a benign technique. Researchers following the 1%-of-body-mass guideline may be collecting too much blood from wild birds, especially when research requires repeated samples over short periods