An image coding method is provided for coding an image on a block-by-block basis. The method includes determining transform unit sizes in a block to be equal to or greater than a preset minimum transform unit size, setting prediction unit sizes in the block to be equal to the determined transform unit sizes, respectively, and predicting the block, on each prediction unit determined by using the prediction unit sizes to generate a prediction block. The method also includes transforming a difference between the block and the prediction block to residual data, on each transform unit determined by using the transform unit sizes, setting coding unit sizes in the block to be equal to the determined transform unit sizes, respectively, and coding the residual data on each coding unit determined by using the coding unit sizes to generate a bitstream.