The compensation effect of control variables in putting was investigated by uncontrolled manifold (UCM) analysis. No one can repeat any movements precisely because of the existence of motor redundancy which enable to achieve the same objective with different set of control variables. In UCM analysis, to quantify the compensation effect of control variables, the valiance of control variables was decomposed into two orthogonal components, one that affects the value of objective variables (VORT) and another that maintain its value (VUCM). Ten participants were asked to hit the 3 and 5 m putts twenty times respectively and a motion capture system (Eagle, Motion Analysis corp.) was used to measure the motions. Relative positions between joints were used to explain the position of the clubhead and to carry out UCM analysis. The result of ANOVA showed that the ratio (VUCM/VORT) at the top of the swing was lower than those at start and impact (p < 0.01). It was also revealed that some control variables have larger VUCM across the movements (p < 0.01) and another set of them has larger VORT around the top of the swing (p < 0.05). These findings suggest that golfers need to restrain only the some of the variability of control variables.