The proposition that variable practice may be superior to constant practice even for consistent transfer situations was tested on a prototypical consistently performed skill, the basketball free throw. 94 participants were matched on free-throw shooting, then randomly assigned to one of four practice conditions, a Constant condition, i.e., at the free-throw line, and three Variable conditions. Under supervision participants practiced shooting free throws four days a week for three weeks. Three substantially different variable practice conditions produced significant improvement similar to that of constant practice on tests during each week of practice and on a delayed retention test. Consistent with the proposition, the most variable practice group performed as well as the other groups on the retention test despite lower practice performance.