Activity monitors such as the SenseWear Pro3 (SWP3) and the activPAL3 Micro (aP(3)M) are regularly used by researchers and practitioners to provide estimates of the metabolic cost (METs) of activities in free-living settings. The purpose of this study is to examine the accuracy of the MET predictions from the SWP3 and the aP(3)M compared to the criterion standard MET values from indirect calorimetry.Fifty-six participants (mean age: 39.9 (+/- 11.5), 25M/31F) performed eight activities (four daily living, three ambulatory and one cycling), while simultaneously wearing a SWP3, aP(3)M and the Cosmed K4B(2) (K4B(2)) mobile metabolic unit. Paired samples T-tests were used to examine differences between device predicted METs and criterion METs. Bland-Altman plots were constructed to examine the mean bias and limits of agreement for predicted METs compared to criterion METs.SWP3 predicted MET values were significantly different from the K4B(2) for each activity (p <= 0.004), excluding sweeping (p = 0.122). aP(3)M predicted MET values were significantly different (p < 0.001) from the K4B(2) for each activity. When examining the activities collectively, both devices underestimated activity intensity (0.20 METs (SWP3), 0.95 METs (aP(3)M)). The greatest mean bias for the SWP3 was for cycling (-3.25 METs), with jogging (-5.16 METs) producing the greatest mean bias for the aP(3)M.All of the activities (excluding SWP3 sweeping) were significantly different from the criterion measure. Although the SWP3 predicted METs are more accurate than their aP(3)M equivalent, the predicted MET values from both devices are significantly different from the criterion measure for the majority of activities.