China General Administration of Sport has published and adopted the National Physical Fitness Measurement (NPFM—preschool children version) since 2000. However, studies on intraday reliability, sensitivity, and minimum detectable change (MDC) are lacking. This study aimed to investigate and compare the reliability, sensitivity, and MDC values of NPFM in preschool children between the ages of 3.5 to 6 years. Six items of NPFM including 10-m shuttle run, standing long jump, balance beam walking, sit-and-reach, tennis throwing, and double-leg timed hop, were tested for 209 Chinese kindergarten children in Beijing in the morning. Intraday relative reliability was tested using intraclass correlation coefficient (ICC3,1) with a 95% confidence interval while absolute reliability was expressed in standard error of measurement (SEM) and percentage of coefficient of variation (CV%). Test sensitivity was assessed by comparing the smallest worthwhile change (SWC) with SEM, while MDC values with 95% confidence interval (MDC95) were established. Measurements in most groups, except 10-m shuttle run test (ICC3,1: 0.56 to 0.74 [moderate]) in the 3.5 to 5.5-year-old groups, balance beam test in 4- and 5-year-old (ICC3,1: 0.33 to 0.35 [poor]) and 5.5-year-old (ICC3,1 = 0.68 [moderate]) groups, and double-leg timed hop test (ICC3,1 = 0.67 [moderate]) in the 4.5-year-old group, demonstrated good to excellent relative reliability (ICC3,1: 0.77 to 0.97). The balance beam walking test showed poor absolute reliability in all the groups (SEM%: 11.76 to 22.28 and CV%: 15.40 to 24.78). Both standing long jump and sit-and-reach tests demonstrated good sensitivity (SWC > SEM) in all subjects group, boys, and girls. Pairwise comparison revealed systematic bias with significantly better performance in the second trial (p<0.01) of all the tests with moderate to large effect size.