Gates installed to protect bats at caves and mines have appeared to be more harmful than helpful, but few studies have quantitatively measured the direct impact of gates on bats. To assess bats' responses to gates during pre-hibernation swarming behavior, we recorded flight speeds, flight behavior, and vocalizations at 28 sites (16 with and 12 without gates) from Ontario to Tennessee. We recorded data at unobstructed entrances and passages, at sites with metal gates already in place, and at sites with newly erected mock wooden gates in place. Bats always circled and retreated more often and passed directly through less often at gates (most at mock gates). Higher bat activity and smaller gate size corresponded with more fly-retreat and circling behavior. Echolocation, communicative calls, and flight speed did not differ consistently as a function of gate presence or absence. Bat flight behavior did not differ based on spacing of vertical gate supports, number of entrances, or gate position (entrance or passage), although bats generally circled more and passed directly through less often in passages. To minimize impacts on bats, gates should 1) be erected in large areas, 2) have a bat chute or open top, 3) be placed at entrances (unless predation is a problem), 4) be placed on flat ground (not on an incline), and 5) be erected gradually over a period of several weeks or months when feasible.