This study was designed to examine the underlying structure of the Test of Gross Motor Development (TGMD) in Ulrich (1985). The TGMD was administered to 644 children who were randomly divided into two groups (calibration group and validation group). The calibration group (n = 324) included 150 boys and 174 girls, and the validation group included 160 boys and 160 girls, ranging from 3 to 10 years. A two-factor model was postulated and supported. According to the model, seven variables measuring children's ability for moving into space loaded on one factor (locomotor skills), while five variables measuring children's ability for controlling objects loaded on the other factor (object control skills). In addition, the proposed model was found to be invariant across the two groups. Good cross-generalizability of the TGMD appears to support its validity. Physical educators working with young children may use it with confidence when assessing and planning physical education programs involving locomotor and object control skills.