A method and a device for transmitting metadata in a WFD are provided. Particularly, a WFD source forms a WFD frame including a WFD header and metadata. The WFD source transmits, to a WFD sink, a packet stream created by packetizing a metadata stream including the WFD frame. The WFD header includes a metadata structure field and a metadata update indicator. The metadata structure field indicates whether the metadata has a video metadata structure or an audio metadata structure. The metadata update indicator indicates that previous metadata is updated such that all of the previous metadata is replaced with new metadata, or that the previous metadata is incrementally updated together with updated metadata.