A number of previous reviews have very eloquently summarized pain models and endpoints in animals. Many of these reviews also discuss how animal models have enhanced our understanding of pain mechanisms and make forward-looking statements as to our proximity to the development of effective mechanism-based treatments. While a number of reports cite failures of animal pain models to predict efficacy in humans, few have actually analyzed where these models have been successful. This review gives a brief overview of those successes, both backward, providing validation of the models, and forward, predicting clinical efficacy. While the largest dataset is presented on treatments for neuropathic pain, this review also discusses acute and inflammatory pain models. Key to prediction of clinical efficacy is a lack of side effects, which may incorrectly suggest efficacy in animals and an understanding of how pharmacokinetic parameters translate from animals to man. As such, this review focuses on a description of the pharmacokinetic-pharmacodynamic relationship for a number of pain treatments that are effective in both animals and humans. Finally we discuss where and why animal pain models have failed and summarize improvements to pain models that should expand and improve their predictive power.