ClotheDreamer is a text-guided clothing generation model based on 3D Gaussian, capable of generating high-fidelity, wearable 3D clothing assets from textual descriptions. It employs a novel representation method called Disentangled Clothe Gaussian Splatting (DCGS), allowing for separate optimization of clothing and the human body. This technology utilizes bidirectional Score Distillation Sampling (SDS) to enhance the quality of clothing and human body rendering and supports custom clothing template input. ClotheDreamer's synthesized 3D clothing can be easily applied to virtual try-on and supports physically accurate animations.