A moving picture decoding method, which generates a predicted image, and having multiple prediction modes including a mode without motion vector decoding, including: selecting a frame(s) to be referenced to in the prediction of each prediction direction in the prediction mode from among multiple candidate reference frames; and selecting motion vector information used in the prediction mode, wherein at the selecting the frame(s) to be referenced to, when candidate blocks adjacent a current block include blocks that use forward reference frames, a forward reference frame having the smallest index number is selected as the forward reference frame for the current block from the forward reference frames that the blocks using the forward reference frames have; and moving picture decoding is performed by generating the predicted image using the information on the selected reference frame and the information on the selected motion vectors in the prediction mode.