Preclinical models that assess "pain" in rodents typically measure increases in behaviors produced by a "pain stimulus." A large literature exists showing that kappa opioid receptor (KOR) agonists can decrease these "pain-stimulated behaviors" following many different pain stimuli. Despite showing apparent antinociceptive properties in these preclinical models, KOR agonists failed as analgesics in clinical trials. Recent studies that assessed decreases in behavior due to a pain stimulus show that KOR agonists are not effective in restoring these "pain-depressed behaviors" to normal levels, which agrees with the lack of effectiveness for KOR agonists in clinical trials. One current explanation for the failure of previous KOR agonists in clinical trials is that those agonists activated beta-arrestin signaling and that KOR agonists with a greater bias for G protein signaling will be more successful. However, neither G protein-biased agonists nor beta-arrestin-biased agonists are very effective in assays of pain-depressed behavior, which suggests that novel biased agonists may still not be effective analgesics. This review provides a concise account of the effectiveness of KOR agonists in preclinical models of pain-stimulated and pain-depressed behaviors following the administration of different pain stimuli. Based on the previous results, it may be appropriate to include both behaviors when testing the analgesic potential of KOR agonists.