In many important search tasks (e.g. medical and baggage screening) targets are rare. Previously, we have demonstrated that miss error rates are 2-3 times higher when targets are rare (1-2% target prevalence) than when common (50% prevalence; Wolfe et al, Nature, 2005). In signal detection terms, prevalence effects reflect shifts in criterion, rather than loss of sensitivity (Wolfe et al., JEP-General, 2007). At low prevalence, observers make more misses and fewer false alarms, and make faster correct rejection responses. In tasks where misses are much less desirable than false alarms, it would be advantageous to counteract the prevalence effect by coaxing observers into adopting the criterion characteristic of high prevalence at low prevalence. Can we shift criterion and "cure" the prevalence effect using a monetary payoff matrix (e.g. Navalpakkam et al, 2007)? Here, using a realistic x-ray baggage-screening task, we compared two payoff matrices. One was biased toward target-present responses: hits $+0.32; false alarms $-0.02; misses $-0.65; correct rejections $+0.01. The other was balanced: $+0.01 for any correct response; $-0.01 for any error. X-ray images of empty bags were "packed" with overlapping x-ray images of weapons and non-weapon objects. Bags contained 3, 6, 12, or 18 objects. Observers searched for guns and knives. The balanced payoff matrix produced the usual prevalence effect (29% misses at low prevalence, 18% at high). The payoff matrix favoring hits modestly shifted criterion toward target present responses and slowed correct target absent responses when targets were rare. However, the reduction in miss errors was disappointing. Observers still made 27% misses at low prevalence. Prevalence effects produced under these conditions remain stubbornly resistant to standard countermeasures.