Salticid spiders are a tractable group for studies of learning. We presented Phidippus princeps Peckham & Peckham 1883 with the challenging task of associating prey with color cues in a T-maze. Experimental spiders were given the opportunity to learn that a cricket was hidden behind a block of a particular color. To eliminate the use of other cues, we randomly assigned both block position within the maze, and maze location within the room. For control spiders, no cues predicted the location of prey. We gave spiders two blocks of trials. Each block consisted of four training trials followed by a probe trial in which no prey was present. Trials lasted an hour, and spiders were given one trial per day. Not all spiders were successful in finding the prey during training trials. In the first probe trial, there was no evidence of learning: there was no effect of treatment, the number of successful training trials, or their interaction on which block the spiders chose first. In the second probe trial, there was a significant interaction between treatment and number of successful training trials: experimental-group spiders with a greater number of successful training trials were more likely to choose the correct block in the probe trial. This study demonstrates that P. princeps can learn the location of prey by color cues alone, a challenging task, and adds to the growing literature on learning in spiders.