Previous research from Rajsic et al. (2015, 2017) suggests that a visual form of confirmation bias arises during visual search for simple stimuli, under certain conditions, wherein people are biased to seek stimuli matching an initial cue color even when this strategy is not optimal. Furthermore, recent research from our lab suggests that varying the prevalence of cue-colored targets does not attenuate the visual confirmation bias, although people still fail to detect rare targets regardless of whether they match the initial cue (Walenchok et al. under review). The present investigation examines the boundary conditions of the visual confirmation bias under conditions of equal, low, and high cued-target frequency. Across experiments, I found that: (1) People are strongly susceptible to the low-prevalence effect, often failing to detect rare targets regardless of whether they match the cue (Wolfe et al., 2005). (2) However, they are still biased to seek cue-colored stimuli, even when such targets are rare. (3) Regardless of target prevalence, people employ strategies when search is made sufficiently burdensome with distributed items and large search sets. These results further support previous findings that the low-prevalence effect arises from a failure to perceive rare items (Hout et al., 2015), while visual confirmation bias is a bias of attentional guidance (Rajsic et al., 2015, 2017).