Acceptance testing is a final stage of testing that is performed on a system prior to the system being delivered to a live environment. Systems subjected to acceptance testing might include such deliverables as a software system or a mechanical hardware system. Acceptance tests are generally performed as "black box" tests. Black box testing means that the tester uses specified inputs into the system and verifies that the resulting outputs are correct, without knowledge of the system's internal workings.

User acceptance testing (UAT), is the term used when the acceptance tests are performed by the person or persons who will be using the live system once it is delivered. If the system is being built or developed by an external supplier, this is sometimes called customer acceptance testing (CAT). The UAT or CAT acts as a final confirmation that the system is ready for go-live. A successful acceptance test at this stage may be a contractual requirement prior to the system being signed off by the client.

In some industries, such as manufacturing, factory acceptance testing (FAT), is the term usually used for the formal acceptance test stage. A factory acceptance test may be performed by the vendor or supplier of the system and observed by the client. This is often done in a very formal manner, with each factory or site acceptance test being documented in detail, and each portion of the test signed off.

Acceptance testing is often made up of many individual "test cases," which are individually documented test scenarios. Each test case that forms part of the overall acceptance test will usually document what the input to the system consists of, and what the expected output should be. If the actual output is as expected, the test case is said to return a successful, or positive result.

When acceptance tests are designed and specified, often a client or end user group will document what level of results will be considered sufficiently successful for the system to be accepted. Payment of the supplier or vendor may depend on this target being met. During acceptance testing, each test case is usually assigned a "pass" or "fail" result. Some test cases may be extremely important, and other test cases may test non-essential parts of the system. As such, for the system to overall achieve a "pass" in acceptance testing, it may not be necessary for every single test case to achieve a positive result.