For the RDBACK processing method, the system uses the data set's trailer labels as header labels, and vice versa. Each label group is read in the normal sequence, that is, EOF1 before EOF2, and so forth. The data records, however, are read in reverse sequence.

Multivolume data sets can be read backward. Concatenated data sets, Format-D (variable-length) records, and Format-S (spanned-length) records cannot be read backward.