Pedometers are used in many studies because physical activities can be simply assessed using them. In addition to the spring-levered type, piezo-electric pedometers have recently been used. However, their accuracy in children has not been investigated in previous studies. In this study, we investigated the accuracy of spring-levered pedometer and piezo-electric pedometer step counts in comparison with hand-tallied step counts with children during self-paced walking. First to 6th-grade primary school children (394 subjects; 201 boys and 193 girls) walked with pedometers on an outdoor 50-m course, and the measured values were compared with actual steps at 3 speeds: normal, slow, and fast paces. The counts obtained by the spring-levered pedometer were significantly lower than the actual steps in most grades at all walking paces. In the normal- and fast-pace walking, the counts by spring-levered pedometer were significantly lower than the actual steps in all grades excluding the 5th grade. Moreover, the spring-levered pedometer underestimated by 25% or more in all grades at slow pace. In contrast, the percent error of the counts by the piezo-electric pedometers at normal pace were mostly within ±3%, confirming their high accuracy. Based on these findings, spring-levered pedometers are not appropriate for children, whereas piezo-electric pedometers are useful for investigation of the physical activity of children.