The literature related to Borg's ratings of perceived exertion (RPE) scale has revealed inconsistencies about the strength of the relationship between ratings of perceived exertion and various physiological criterion measures, such as heart rate, blood lactate concentration, percent maximal oxygen uptake (%VO2max), oxygen uptake (VO2), ventilation and respiration rate. Using sex of participants, fitness, type of RPE scale used, type of exercise, exercise protocol, RPE mode and study quality, we undertook a meta-analysis to determine the strength of the relationship between RPE scores and the six aforementioned physiological measures. The weighted mean validity coefficients were 0.62 for heart rate, 0.57 for blood lactate, 0.64 for %VO2max 0.63 for VO2, 0.61 for ventilation and 0.72 for respiration rate. Analysis of moderator variables revealed that the following study features could account for the variation of results across studies: heart rate--fitness, type of exercise, protocol and RPE mode; blood lactate concentration--sex, RPE scale; VO2--sex, exercise type, RPE mode; ventilation--sex, RPE mode; respiration rate--exercise protocol, RPE mode. The highest correlations between ratings of perceived exertion and the various physiological criterion measures were found in the following conditions: when male participants (whose VO2 or ventilation was measured) were required to maximally exert themselves (measuring %VO2max or ventilation); when the exercise task was unusual [e.g. when participants were swimming, which is less common than walking or running (when heart rate, %VO2max and VO2 are measured)]; or when the 15-point RPE scale (measuring blood lactate concentration) was used. These findings suggest that although Borg's RPE scale has been shown to be a valid measure of exercise intensity, its validity may not be as high as previously thought (r = 0.80-0.90), except under certain conditions.