In order to minimize the bias of Constant score we modified the allocation of strength subscore. One hundred and two patients with 3- and 4-part proximal humerus fractures were treated using locking plate fixation and followed up for > 1 year. The clinical outcomes were assessed by DASH score abbreviated Constant score (AbbCS strength item excluded) modified Constant score (ModCS with 12-pound strength) and original Constant score (CS with 25-pound strength). The satisfaction rate was determined for each scoring instrument. Compared to CS the satisfaction rate was significantly higher in DASH score AbbCS and ModCS (all p < 0.001) but the latter 3 groups did not show significant difference. ROC analysis showed that a > 7-pound shoulder strength was present in patients with satisfied outcome assessed by DASH score. In conclusion strength evaluation is necessary for the assessment of shoulder function but the over-allocated strength should be modified in Constant score.