Background: School-based interventions are a key opportunity to improve children’s physical activity (PA); however, there is lack of evidence about how pedagogical approaches to motor learning in physical education (PE) might affect PA in children. Therefore, this study aimed to assess how different pedagogical approaches in PE might affect children’s PA. Methods: Participants (n = 360, 5–6 years) from 12 primary schools within the SAMPLE-PE randomized controlled trial were randomly allocated to either Linear Pedagogy (LP: n = 3) or Nonlinear Pedagogy (NP: n = 3) interventions, where schools received a 15-week PE intervention delivered by trained coaches, or to a control group (n = 6), where schools followed usual practice. ActiGraph GT9X accelerometers were used to assess PA metrics (moderate-to-vigorous PA, mean raw acceleration and lowest acceleration over the most active hour and half hour) over whole and segmented weeks at baseline, immediately post-intervention and 6 months follow-up. Intention to treat analysis employing multilevel modelling was used to assess intervention effects. Results: LP and NP interventions did not significantly affect children’s PA levels compared to the control group. Conclusion: PE interventions based on LP and NP alone might not be effective in improving habitual PA in children.