The document is a master's thesis titled "Automatic Program Parallelization for GPU Environment". It discusses using a compiler called C2CUDA to automatically parallelize sequential C programs for execution on a GPU. The thesis presents techniques for data dependence analysis and loop transformations to expose parallelism. It also provides results of experiments parallelizing matrix operations and magnetic resonance imaging algorithms using C2CUDA. Speedups of up to 58x were achieved for the parallelized applications compared to sequential execution.