Mask-based lensless cameras can be flat, thin and light-weight, which makes them suitable for novel designs of computational imaging systems with large surface areas and arbitrary shapes. Despite recent progress in lensless cameras, the quality of images recovered from the lensless cameras is often poor due to the ill-conditioning of the underlying measurement system. In this paper, we propose to use coded illumination to improve the quality of images reconstructed with lensless cameras. In our imaging model, the scene/object is illuminated by multiple coded illumination patterns as the lensless camera records sensor measurements. We designed and tested a number of illumination patterns and observed that shifting dots (and related orthogonal) patterns provide the best overall performance. We propose a fast and low-complexity recovery algorithm that exploits the separability and block-diagonal structure in our system. We present simulation results and hardware experiment results to demonstrate that our proposed method can significantly improve the reconstruction quality.