When a test is revised, it is important that test users are made aware of the comparability of scores of the new and the original test. We examined how scores on the Test of Gross Motor Development–Second Edition (TGMD-2) and Test of Gross Motor Development–Third Edition (TGMD-3) compared among children in middle childhood. Participants were 270 children recruited in grade 3 (54% female; M age = 8 years 6 months) and followed through grade 5. Participants completed the skills of both tests. Subtest scores were converted into percent of maximum possible (POMP) scores to facilitate comparison. Although similar, uniformly the TGMD-3 POMP scores were slightly lower. Repeated measures analyses of variance revealed that locomotor subtest scores derived from both tests improved from grade 3 to grade 5, as did TGMD-3 assessed ball skills. However, there was no difference in TGMD-2 assessed object control skills over time. It appears that under-contribution by the underhand roll suppressed the trajectory of improvement of TGMD-2 assessed object control skills. This finding supports the exclusion of the roll from the TGMD-3. The consistent pattern of sex-based differences in TGMD-2 object control skill and TGMD-3 ball skills reinforces the need for male and female norm-reference data for ball skills.