1. Originally, a method of organizing work for a computer system, designed to reduce overheads by grouping together similar jobs. One form avoided reloading systems software. The jobs were collected into batches, each batch requiring a particular compiler, the compiler was loaded once, and then the jobs submitted in sequence to the compiler. If a job failed to compile it took no further part in the processing, but those jobs that did compile led to the production on magnetic tape or other backing store of an executable binary. At the end of the batch of compilations those jobs that had produced an executable binary form were loaded in sequence and their data presented to the jobs. Another form avoided the time taken to read cards and print on paper by offline processing, having a batch of jobs on magnetic tape.

The term has also come to be applied to the background processing of jobs not requiring intervention by the user, which takes place on many multiaccess systems.

2. A method of organizing a data processing system in which transactions are input in a batch, sorted, and sequentially processed to update and/or query a master file. This is the only possible method if magnetic tape is used as backing store; there are applications where it is the most efficient method even using disks. See also transaction processing.