Each test that is executed consists of at least one step: the "primary" step. During execution, each test may spawn additional nested steps that may run in parallel or in series with one another as required.

A test step may be used to describe a particular instance of a data driven test. It can also be used to generate a completely dynamic hierarchy at runtime that mirrors and extends the static test hierarchy with new information derived from the actual execution of the test.

Each step has its own execution log and test result (pass/fail/inconclusive). Therefore a multi-step test may possess multiple execution logs and test results. This is deliberate. Think of a TestStep as being the declarative component of a test that specifies which test method to invoke and its arguments. A Test is the runtime counterpart of the Test that captures output and control flow information about part or all of the test.

A test step also has metadata that can be update at run-time to carry additional declarative information about the step.