Provided are a method and apparatus for performing combination prediction by obtaining a weight for intra prediction and inter prediction by taking into account at least one of a distance between a reference picture and a current picture, a size of a current block, and characteristics of the inter prediction and intra prediction. An image decoding method according to the present disclosure includes parsing, from a bitstream, combination prediction information indicating whether to predict a current block by combining intra prediction and inter prediction; determining whether to perform combination prediction on the current block, based on the combination prediction information; obtaining, when the combination prediction is performed, a first prediction value by performing the inter prediction on the current block, and obtaining a second prediction value by performing the intra prediction on the current block; determining a weight for the inter prediction and a weight for the intra prediction, based on at least one of a distance between a reference picture and a current picture, a size of the current block, and characteristics of the inter prediction and the intra prediction; and performing the combination prediction, based on the weight for the inter prediction, the weight for the intra prediction, the first prediction value, and the second prediction value.