Striped patterns are common in nature and are used both as warning signals and camouflage. Their effectiveness in either role depends on their color and spatial frequency, and how these compare to the background. Although this general principle is well established, the specific detail of how visual texture influences defensive coloration remains untested in the field. For aposematic patterns, especially, little work has focused on how pattern components, as opposed to color, affect warning signal efficacy. By presenting artificial moth-like stimuli, pinned to tree bark, to wild avian predators, and human observers, we examine how the spatial frequency and orientation of stripes affects the survival and detectability of yellow-and-black (aversive) and olive-and-black (cryptic) patterns. For the cryptic stripes, we find that matching the dominant spatial frequency and orientation of the background increases survival against bird predation and decreases the distance from which humans first detect the target. For aversive stripes, however, survival against birds peaked at spatial frequencies that neither matched the dominant background spatial frequency nor maximized the mismatch between target and background. This peak in survival at intermediate spatial frequencies did not match detectability by humans: there was no difference in the initial detection distance between stripes of different spatial frequencies, although the distance at which stripes could be resolved did differ. We suggest that, although the best cryptic strategy is to match the dominant components of the background as closely as possible, the optimal aposematic signal is one that balances signal distinctiveness and recognition at a distance.