Hickey, Chelazzi & Theeuwes (2010; 2011) have reported an intriguing effect of reward upon visual search performance. In an odd-one-out paradigm a diamond target appears in an array of circles (or vice versa). A salient color singleton is presented on some trials, while on other trials all items are uniformly colored. Observers are rewarded for correct responses with randomly varying ‘high’ or ‘low’ monetary values. Since the reward is random, it is not contingent upon any aspects of the display. Hickey et al. found that responses to targets following ‘high’ reward trials are speeded when the color scheme of the previous trial is repeated while a color scheme swap resulted in an inhibitory effect. In two experiments, we tried to expand on these findings. We tested whether the results would generalize to identical contingencies and variables, but different stimulus materials (colored Gabors, rather than shapes). Observers were tested on singleton versus no-singleton trials varying within blocks or run in separate blocks. Reward had no effect on performance at all. Observers might as well have ignored reward altogether. This was true of both mixed and blocked conditions. This failure to replicate does not necessarily come as a surprise, however. The boundary conditions of non-contingent reward priming may be very subtle and depend heavily on an observer’s attitude towards the task. A corollary of a non-contingent reward schemes is that the strategy to ensure maximum monetary pay-off is to maximize accuracy in any condition, which may overshadow beneficial and/or inhibitory effects of reward values. We discuss how task contingent reward schemes may be better suited to reveal the functional interplay of reward and visual attention.