The best-supported hypothesis for why zebras have stripes is that stripes repel biting flies. While this effect is well-established, the mechanism behind it remains elusive. Myriad hypotheses have been suggested, but few experiments have helped narrow the field of possible explanations. In addition, the complex visual features of real zebra pelage and the natural range of stripe widths have been largely left out of experimental designs. In paired-choice field experiments in a Kenyan savannah, we found that hungry Stomoxys flies released in an enclosure strongly preferred to land on uniform tan impala pelts over striped zebra pelts but exhibited no preference between the pelts of the zebra species with the widest stripes and the narrowest stripes. Our findings confirm that zebra stripes repel biting flies under naturalistic conditions and do so at close range (suggesting that several of the mechanisms hypothesized to operate at a distance are unnecessary for the fly-repulsion effect) but indicate that interspecific variation in stripe width is associated with selection pressures other than biting flies.