The purpose of this study was to investigate college volleyball coaches’ leadership behavior and its relationship with team cohesion. The participants of the study were 121 volleyball players at the first level of the male and female sections of the college volleyball league contests it the academic year 2001. The research measurements were Leadership Scale for Athletic Preference and Athletic Team cohesion. The effective questionnaires were 206 copies and analyzed by the following analysis methods: factor analysis, descriptive statistics, test of independence, one-way variance, stepwise regression analysis and Pearson product moment correlation. The results showed that there were significant differences in volleyball players’ preferences in following dimensions: (a)There were obvious differences shown in preference for coaches’ autocratic behaviors from different gender players’ the preference for coaches’ leadership behaviors. (b)There were obvious differences shown in the perceptions of autocratic behaviors from different training frequency players’recognitions of coaches’ leadership behaviors. (c)There were obvious differences shown in preference for democratic behaviors and autocratic behaviors from different athletic achievement players’ recognitions of coaches’leadership behaviors. (d)Coaches’ leadership behavior can effectively predict team cohesion with respect to the perceptions of coaches’leadership behaviors concerning training and instruction, social support, positive feedback and democratic behavior, and preference for coaches’ leadership behaviors concerning positive feedback, democratic behavior and social support. (e)The autocratic behavior factor of preference for coaches’leadership behaviors in predicating team cohesion has the 8% explanatory amount for social cohesion. The positive feedback factor of coaches’leadership behaviors in predicting team cohesion has 9% explanatory amount for work cohesion.