In kinesiology, many data are compositional (percentage) data. For example, the daily food consumption is compositional data that is a composition of fat, protein, and carbohydrates. A physical education teacher’s time spent in class can be categorized as the percentage of time spent on instruction, management, and observation. The intrinsic characteristic of the compositional data is that the proportions in a composition are subject to a unit-sum constraint. Literature has shown that the general approach (e.g., ANOVA), which is designed to analyze un-compositional data, should not be used directly to analyze compositional or percentage data. John Aitchison (1986) developed a new methodological approach for the statistical analysis of compositional data. However, one of the difficulties for appropriately analyzing compositional data in kinesiology is that there is no comprehensive approach to help practitioners appropriately apply Aithison’s new methodology. The purpose of this project was to systematically introduce a practical approach to apply Aitchison’s methodology in the analysis of compositional data. Using empirical compositional data in the field of kinesiology, the descriptive and inferential statistics on investigating the group differences was introduced. The step-by-step procedure for conducting compositional data analysis included: (1) Ternary diagrams for the description of the integrative variability of the compositional data, (2) Relative descriptive statistics including relative means and relative variances, and (3) Compositional data analysis on the group comparisons. Limitations of the existing methods and future research direction are also discussed.Keyword(s): assessment, measurement/evaluation, research