Procedural abstraction is the process of converting a specific procedure into a general procedure by ignoring certain details. This is used during the design process to allow the programmer to focus on the structure of the program instead of the detail of the individual functions.

A procedural abstraction specifies everything the users require from the procedure but nothing more. All other details are left up to the programmer to determine during the implementation. Procedural abstractions are sometimes referred to as "black boxes" since they describe what a procedure does without describing how it does it.

Defining a procedural abstraction requires two steps. The first step names the input and output parameters as well as their types. The second step defines the conditions (called the requires clause), any side effects (the modifies clause) and what the abstraction achieves (the effects clause).

When a procedural abstraction is defined completely, it has the properties of locality and modifiability. Locality means that the details of the implementation are local to the individual procedure and only need to be known by the programmer dealing with the implementation. Modifiability means that replacing the implementation of the procedure does not affect the rest of the program.