It is well known that attentional selection is influenced by our previous experience of rewards. Stimuli whose selection was previously rewarded continue to capture attention in a subsequent test session when these rewards are no longer available. In this study we provide evidence of both attentional and oculomotor capture by stimuli that merely signal the magnitude reward available on a particular trial. The selection of these stimuli was not necessary but rather detrimental for actual payout. Participants searched for a target (a diamond) among 5 uniquely colored circles. The color of one of these circles signaled the magnitude of reward available for that trial (e.g., a red circle would indicate a high reward, a green circle a low reward). Even though this colored circle signaling the reward availability was never part of the task set nor physically salient, it captured both attention and the eyes. Our results suggest that stimuli signaling reward get prioritized through reward-learning and that this learning only occurs when the reward signaling stimuli are attended. We conclude that task-irrelevant and non-salient stimuli that signal the availability of reward gain priority in attentional selection even if selecting them is not necessary but rather detrimental for reward payout. Furthermore we show that a stimulus signaling reward only gained priority if it was the only stimulus predicting reward outcome. These findings expand the growing evidence which suggests that attentional selection cannot fully be explained in terms of the traditional separation in top-down and bottom-up processes (Awh, Belopolsky & Theeuwes, 2012).