If format-F records with keys are being written, the key of each record can be used to identify the record. For example, a data set with keys ranging from 0 to 4999 should be allocated space for 5000 records. Each key relates directly to a location that you can refer to as a relative record number. Therefore, each record should be assigned a unique key.

If identical keys are used, it is possible, during periods of high processor and channel activity, to skip the desired record and retrieve the next record on the track. The main disadvantage of this type of organization is that records might not exist for many of the keys, even though space has been reserved for them.