Motor schemas are very attractive concepts; however, their development and internal structure appear difficult to probe and identify. The present study focuses on the development of a motor schema in constant practice. In four experiments, participants practiced bimanual juggling with three standard balls. During retention tests they were asked to juggle with standard ("practiced") balls and nonstandard ("nonpracticed") balls of different weight, shape, length, and size. Four juggling experiments were carried out. The participants achieved the best results juggling with the practiced balls in three experiments. During one experiment, they achieved better results juggling with nonpracticed balls. However, in all four experiments, a correlation was observed between the number of correct juggling sequences with standard balls and with nonpracticed balls. Scores presented as mean numbers of juggling cycles performed in retention tests revealed a gradient with its peak value for juggling with the practiced balls. In one experiment, the hypothetical gradient slope was steeper; however, a strong correlation between scores for juggling with practiced and nonpracticed balls was noted.