The aim of this study is to investigate the relationships among segmental accuracy, suprasegmental accuracy (intonation), fluency of speech rate/pause, intelligibility and comprehensibility criterions in pronunciation assessment. These criterions are widely regarded as goals for second language teaching and testing for pronunciation. However, there is little empirical evidence about what these criterions are measuring in terms of actual pronunciation skills and how they influence each other or even whether they are related. So in this study, based on previous research, a Path Model was developed and evaluated to illustrate relationships among these criterions. The study found that there are significant strong correlations among accuracy, fluency, intelligibility and comprehensibility. But there was no direct effect found between suprasegmental accuracy and intelligibility or between suprasegmental accuracy and comprehensibility. However there was a direct effect found for both segmental accuracy and fluency of speech rate/pause on both intelligibility and comprehensibility. First, regression analysis revealed intelligibility accounted for R-squared 97% of variability in pronunciation scores for these two criterion. These two criterion, segmental accuracy(.51) and fluency of seech rate/pause(.50), related to intelligibility to a similar degree. Second, regression analysis showed comprehensibility accounted for R-squared 98% of variability in pronunciation scores for these two criterion. Fluency of speech rate/pause(.67) related to comprehensibility relatively higher than segmental accuracy(.18). Intelligibility was also related to comprehensibility but to a small degree(.17) in regards to semantic understanding. (Ewha Womans University)