Abstract Background When patients suffering from fibromyalgia undergo a therapeutic trial, a non‐negligible part of analgesia is not explained by the drug itself. The mechanisms of this non‐specific effect need to be understood. Materials and methods We undertook secondary analyses of a double‐blind randomized trial in fibromyalgia patients in which 100 mg/day milnacipran was not found superior to placebo. Data from 49 patients belonging to both groups were pooled. Both before treatment and one month after treatment, all patients underwent a CaNTAB ® neuropsychological test (related to spatial planning, reaction time, decision‐making and risk‐taking, and ability to name objects), and measurements of sensation and pain thresholds to heat and cold, supraliminal heat pain threshold, punctuate mechanical pain threshold and temporal summation, mechanical allodynia to skin brushing, and response to conditioned pain modulation. We studied the baseline predictors of analgesia and the indicators of change associated to analgesia separately. A stepwise approach was used to select the factors to enter into the final ANCOVAs, in which age, body mass index, treatment group and pain at baseline were covariates. Results No baseline predictor of non‐specific analgesia other than pain at baseline was found to be predictive. Conversely, several neuropsychological (higher performance) or psychophysical (lower sensitivity) changes correlated with analgesia in unadjusted analyses. Multivariable analyses identified increases in warm/heat thermal thresholds and an increased ability to name objects, as factors associated with analgesia. Conclusions The changes observed concomitantly to non‐specific pain analgesia might be related to mild changes in brain functioning, based on convergent literature data.