Purpose: Chronic pain is the primary complaint associated with OA, serving as a predictor of physical dysfunction and muscular weakness. As blinded, randomized placebo-controlled trials of therapeutic agents in OA have repeatedly demonstrated a marked reduction in pain response to placebo, studies of the brain response to placebo may inform on brain properites of OA itself. Although mechanisms of placebo response have been repeatedly studied in humans, we are the first to investigate its underlying brain properties in a clinical drug study design. Here we investigate the possibility of identifying brain properties that predispose different OA patients to placebo analgesia. Methods: 20 patients meeting ACR criteria for knee OA were enrolled into this single-blinded 4 week study. At baseline, patients were instructed that they might be receiving either active drug or placebo; however they all were given placebo for 2 weeks after which the treatment was stopped. At baseline and at the end of the 2 weeks the patients underwent an fMRI brain scan as well as a structural high resolution T1 brain scan; efficacy reports were collected on both occasions. Two weeks post cessation of medication patients were contacted again and same efficacy reports were recollected. The primary efficacy measure was Visual Analogue Scale 24 hour average pain rating. Secondary efficacy measures included WOMAC Index; Pain detect, Beck depression Index (BDI), and Pain catastrophizing scale (PCS). In addition, 17 healthy age- and gender-matched participants without knee pain were scanned at baseline to serve as controls. For the analysis, the patient groups were stratified as responders vs. non-responders to the placebo treatment based on a decrease in visual analogue scale (VAS) pain rating equal to or greater than 20% from baseline. Analysis was then conducted to determine differences in brain fMRI data between OA patients and controls as well as between OA placebo-responders vs. OA placebo-non-responders. Results: Of the 20 patients enrolled, 9 were male and 11 female; mean age was 57.8±6.6 years; mean duration of OA was 12.1±10.0 years. 3 patients discontinued prior to their week 2 visit; of the remainder, 8 responded to placebo (mean VAS dropped from 72.5±14.1 to 32.5±20.5, i.e., 56% decrease in VAS) whilst 9 did not show any significant change in reported pain (VAS changed from 66.7±8.2 to 70±4.7, i.e., a mean increase of 7%). OA vs controls: By comparing the general number of connections across all brain regions between OA patients and controls we conclude that the region constituted by the medial prefrontal cortex (mpfc) and nucleus accumbans as well as the caudate is more connected (p <0.05, corrected for multiplicity) in patients with OA than in controls. OA-responders vs. OA-non responders: The secondary somatosensory cortex (S2) was generally more linked in the OA non-responders (p-corrected < 0.05). In this group, the S2 region displayed significantly more connectivity with the ventro-lateral prefrontal cortex (vlpfc). Moreover, the vlpfc showed more connectivity (p-corrected < 0.05) with the part of the brain network identified for OA, namely mpfc. Both S2 counts and vlpfc-mpfc connectivity predicted with very high accuracy (AUC=. 97 and .96, respectively) responders to placebo. Conclusions: In this study, we have shown that brain information sharing (functional connectivity) is different in OA in contrast to healthy control subjects. Furthermore, analysis of activity of components of this network predicts the response to placebo. Thus, placebo analgesia is predictable from brain connectivity (information shared between brain regions), prior to start of the placebo treatment trial. These results suggest that distinct sub-groups of OA have distinct propensities for placebo based on their brain states. The specific characteristics that distinguish between these groups remains to be identified. The results of the present study have important implications regarding OA, clinical trials in OA, and for future designs of treatments for OA.