A data encoding method and apparatus and a data decoding method and apparatus in a passive optical network (PON) system include collecting N data blocks at a physical coding sublayer and generating valid data by combining the N data blocks, generating a payload, where the payload includes the valid data, performing FEC encoding on the payload to generate a check part, and generating a codeword structure. The synchronization header may be located at the head or the tail of the codeword structure.