Volume is generated directly from trading activities. More and more academic studies in finance begin to focus on it to explore various characters of investors' behaviors and to model financial markets. This paper shows that K+1 fund separation theorem implies an approximate linear K factor structure for trading volume. The linear K factor model is empirically tested using principal components analysis on turnover data of all securities in Shanghai and Shenzhen stock market over a 168 week sample period (from Jan．3rd，1 995 to May 1 5，1 998)．Further analyses on first two principal components were conducted 1o investigate implications of trading volume in Chinese stock market．