The method of saving data in chunks (ie, where the data is preceded by an 8 byte header consisting of a 4 char ID and a 32-bit size field) is the basis for Interchange File Format. You should now read the article About Interchange File Format for background information.

As mentioned, MIDI File format is a "broken" IFF. It lacks a file header at the start of the file. One bad thing about this is that a standard IFF parsing routine will choke on a MIDI file (because it will expect the first 12 bytes to be the group ID, filesize, and type ID fields). In order to fix the MIDI File format so that it strictly adheres to IFF, Microsoft simply made up a 12-byte header that is prepended to MIDI files, and thereby came up with the RMID format. An RMID file begins with the group ID (4 ascii chars) of 'R', 'I', 'F', 'F', followed by the 32-bit filesize field, and then the type ID of 'R', 'M', 'I', 'D'. Then, the chunks of a MIDI file follow (ie, the MThd and MTrk chunks). So, somewhere after the first 12 bytes of an RMID file, then you should find an embedded MIDI file (although there may be other "chunks" of data before it. Simply skip over those as you would any unknown chunk).

Note that chunks within a MIDI file are not padded out (with an extra 0 byte) to an even number of bytes. I don't know as if the RMID format corrects this aberration of the MIDI file format too.