Index records are stored in control intervals the same as data records, except that only one index record is stored in a control interval, and there is no free space between the record and the control information. So, there is only one RDF that contains the flag X'00' and the length of the record (a number equal to the length of the control interval minus 7). The CIDF also contains the length of the record (the displacement from the beginning of the control interval to the control information); its second number is 0 (no free space). The contents of the RDF and CIDF are the same for every used control interval in an index. The control interval after the last-used control interval has a CIDF filled with 0s, and is used to represent the software end-of-file (SEOF).

Index control intervals are not grouped into control areas as are data control intervals. When a new index record is required, it is stored in a new control interval at the end of the index data set. As a result, the records of one index level are not segregated from the records of another level, except when the sequence set is separate from the index set. The level of each index record is identified by a field in the index header (see Header Portion).

When an index record is replicated on a track, each copy of the record is identical to the other copies. Replication has no effect on the contents of records.

Figure 1 shows the parts of an index record.

Figure 1. General Format of an Index Record