Accurate measurement of physical activity is critical to understand its role in cardiometabolic health and obesity development in children and to monitor trends in behavior and evaluate interventions. An ongoing mixed-longitudinal study of child growth and development in Samoa is collecting physical activity data with both accelerometers and the Netherlands Physical Activity Questionnaire (NPAQ). The aims of our analyses were to (1) describe the response frequency and correlations of individual questions in the NPAQ, (2) develop modified NPAQ scores with selected questions and (3) examine the concordance of modified NPAQ scores with accelerometer outcomes among children aged 2–4 years. We developed two modified NPAQ scores with combinations of questions and assessed concordance of the modified scores with accelerometer data using estimated marginal means adjusted for monitor wear time. Although the evenly distributed tertiles of the modified 15-point NPAQ score showed promising trends of increasing minutes of accelerometer-assessed high-intensity physical activity with increasing tertile, the estimated marginal means were imprecise with high variance, demonstrating that NPAQ score could not accurately assess physical activity levels of preschool-aged children in Samoa. Considering that questionnaires are often considered more cost-effective tools for physical activity measurement than accelerometry, further research is necessary to develop a culturally and age-appropriate physical activity questionnaire in this population.