Thanks to the recent advance in the multimedia techniques, increasing research attention has been paid to the virtual try-on task, especially with the 2D image modeling. The goal of the traditional try-on task is to generate the person image with a new clothing item naturally aligned with the person body shape. In fact, towards try-on, people may also be interested in their try-on looks with different poses. Therefore, in this work, we define a new challenging try-on task, which allows users to see different try-on effects with the new clothing item in various desired poses. In a sense, different from the typical try-on task, we aim to simultaneously change the clothing item and the pose of the given person image. Accordingly, we propose a pose-guided virtual try-on scheme based on the generative adversarial networks (GAN) with a bi-stage strategy. In particular, in the first stage, we propose a shape enhanced clothing deformation model for warping the clothing item where a body shape mask prediction module is introduced to enhance the performance. For the second stage, we present an attention-based bidirectional generative network (AB-GAN), jointly modeling the attentive clothing-person alignment and the bidirectional generation consistency. For evaluation, we construct a new large-scale FashionTryOn dataset comprising 28,714 triplets with each consisting of a clothing item image and two model images in different poses. Extensive experimental results conducted on our FashionTryOn dataset show the superiority of our model over several state-of-the-art methods.