In the safety critical systems for high reliability, large number of tests are required which is usually impractical for cost and time. To avoid this, fault tree analyses are performed using the failure database handbook, which are however less useful for new development. Simulations of the physical model are also employed but they are inaccurate due to the many uncertainties. To overcome this problem, recent trend is to evaluate the reliability by quantifying the uncertainties in the physical model, which is validated or calibrated by the tests. The system reliability is then obtained by building block integration of simulation and tests. Numerous probabilistic techniques are employed to perform the test and evaluation more efficiently. Application examples can be found in the reliability prediction of rocket system and pyrotechnic actuation devices.