All new physical behaviour measurement devices should be assessed for compatibility with previous devices. Agreement was assessed between the activPAL4TM and activPAL3TM physical behavior monitors within a laboratory and a multi-day free-living context. Healthy children aged 6–12 years performed standardised (sitting, standing, stepping) (12 min) and non-standardised (6 min) activities in a laboratory and a multi-day (median 3 days) free-living assessment whilst wearing both monitors. Agreement was assessed using Bland–Altman plots, sensitivity, and the positive predictive value (PPV). There were 15 children (7M/8F, 8.4 ± 1.8 years old) recruited. For the laboratory-based standardised activities, sitting time, stepping time, and fast walking/jogging step count were all within ±5% agreement. However, the activPAL4TM standing time was lower (−6.4%) and normal speed walking step count higher (+7.8%) than those of the activPAL3TM. For non-standardised activities, a higher step count was recorded by the activPAL4TM (+4.9%). The standardised activity sensitivity and PPV were all >90%, but the non-standardised activity values were lower. For free-living agreement, the standing time was lower (−7.6%) and step count higher (all steps + 2.2%, steps with cadence >100 step/min + 6.6%) for the activPAL4TM than the activPAL3TM. This study highlights differences in outcomes as determined by the activPAL4TM and activPAL3TM, which should be considered when comparing outcomes between studies.