Rationale The extent to which placebo effects can be driven exclusively by subjective impressions of improvement in the absence of any independent corroboration is unclear. Methods Thirty-six self-referred patients were treated with a light therapy device intended to rejuvenate facial skin. At each of eight weekly treatments, participants' facial skin was exposed for 40 seconds to pulses of multispectral LED-generated light in the range of 588 nm wavelength at 0.1 J cm(-2). Outcomes were assessed by participants as well as by the treating doctor and by blinded, expert raters. Results Patients reported robust and statistically significant improvements in seven facial features at the conclusion of the 8-week treatment regimen as well as at 1-month follow-up (for all comparisons, P </= 0.003, median d = 1.14). In sharp contrast, both the treating doctor and blinded, expert raters were unable to detect any improvement whatsoever (for all comparisons, P > 0.05). Moreover, effect sizes were close to zero and in the opposite direction from improvement (median d = -0.06 for doctor ratings; and for observer ratings, there was only a 46% success rate at identifying post-treatment as compared with pre-treatment photographs). Conclusion The robust placebo responses documented in this trial were confined to the subjective impressions of the patients. Neither the treating doctor nor blinded, expert raters could detect any improvement. Thus, patients can perceive improvement in medical interventions in the absence of any independent corroboration that improvement has occurred. This result is used a heuristic to more clearly define the components of the placebo response.