Batch processing is the sequential execution of a series of programs ("jobs") on a computer. The term batch originated in the days when programmers wrote code on paper forms which would be keypunched onto 80-column punch cards or paper tape. The cards or tape would then be passed to the system operator ("sysop") who would schedule the task and later stage feed the program into the computer. Because a number of tasks would be 'batched' together rather than be put into the system immediately the tasks were referred to as "batch jobs". Batch jobs are not thought of as interactive, but transaction processing systems such as IBM's CICS may run from a technical perspective as batch jobs, although this is typically obscured from terminal users.

In many companies, the batch jobs would be scheduled on a timetable, for example 'end of day' and 'end of quarter' and could be initiated automatically by the Job Control Language or manually by the operator. The major benefit of batch processing is controlling the load placed on the computer by careful scheduling. For example, a long processor-intensive task would be timed to run over-night allowing for timesharing usage or short jobs during the day. When a batch job begins, processing will usually continue until it is completed unless there is an error but some mainframe machines could process multiple batched jobs concurrently.

Although the use of batch processing reduced with the move to personal computing and the drop in computing costs it is still used extensively by many large commercial businesses for back office tasks.

The opposite to batch processing is transaction processing and is not to be confused with DOS batch files.