Compressive sensing based on wavelet is applied to ultrasonic medical image. Structured random measurement matrix (SRM) is chosen as measurement matrices to reduce computation. The image is sparsely represented in the wavelet transform space. Finally, Gradient Projection for Sparse Reconstruction (GPSR) is used to reconstruct the image. The experiments show that the ultrasonic medical images could be reconstructed well, which proves our approach is practicable to in the reconstruction of the ultrasonic medical image.