With the rapid increase of personalized garment requirements, intelligent garment design has attracted more and more attention. The study reported in this paper concentrates on the modularized design method of garments in order to provide theoretical support for intelligent garment design. The proposed modularized design method consists of composition and transformation of garment parts as well as their configuration and connection. There are two categories of garment parts: essential parts and non-essential parts, each of which is determined by their key design attributes. Rules of transformation for garment parts were analyzed by classification of transformation rules. Binary decision variables were introduced to decide if garment parts should be selected. A mathematical model was developed to analyze constraints of garment parts connection.