Individual differences in pharmacokinetics and pharmacodynamics, the type of pain and the method of drug administration can account for the response variability to analgesics. By integrating a clinical and an experimental approach, we report here that another important source of variability is represented by individual differences in non-specific (placebo) activation of endogenous opioid systems. In the first part of this study, we analyzed the effectiveness of buprenorphine, tramadol, ketorolac and metamizol in the clinical setting, where the placebo effect was completely eliminated by means of hidden infusions. We found that the hidden injections were significantly less effective and less variable compared with open injections (in full view of the subject), suggesting that part of the response variability was due to non-specific factors (placebo). Since we could not administer the opioid antagonist, naloxone, to these patients, in the second part of this study, we induced experimental ischemic arm pain in healthy volunteers and found that, as occurred in clinical pain, the analgesic response to a hidden injection of the non-opioid ketorolac was less effective and less variable than an open injection. Most importantly, we obtained the same effects by adding naloxone to an open injection of ketorolac, thus blocking the opioid-mediated placebo component of analgesia. These findings indicate that both the psychological (hidden injection) and pharmacological (naloxone) blockade of the placebo response reduce the effectiveness of, and the response variability to, analgesic drugs. Therefore, an important source of response variability to analgesics appears to be due to differences in non-specific activation of endogenous opioid systems.