Based on the research about outside shape of woman warm jacket more than twenty years, fashion variables that were representative and can describe the fashionable shape were discussed. Experiment was designed to achieve data of large numbers of female body. Body size variables were statistically analyzed to decide the module that was the basement for achieving data from historical photos. Fashionable characteristic diagrams of garment length, front chest width, shoulder length, collar height and their error bar charts were drawn for discussing the change of fashionable shape. The fashion trends in the future were also prognosticated scientifically. At the same time, a historical database was developed for manufacture and designing, which it is the basement for automatic pattern designing. This new method for fashion trend research was introduced by data mining technology, which it opens our minds for garment science research and offers a new database for improving garment CAD system.