Abstract: Using the methods of questionnaires and mathematical statistics, the research conducted a stratified random household survey of 3 304 residents in Beijing, to investigate the costs of physical training, sport information, sports events, physical fitness, sports facilities, as well as sports shoes and clothing among the respondents. Then the general linear regression model (OLS) and quantile regression (QR) are used to analyze the status quo and the differences of sport consumption among residents with different characteristics. The results show that the gender, marriage, age, income, regions and types of work have influences on their sport consumption, with the significant differences in 30%, 40%, 50%, 60%, 70%, 80% and 90% quantiles, respectively. The results indicate that quantile regression is more accurate than general linear regression in reflecting the influence of different distribution of independent variables on dependent variables.