The reliability tests play more and more important role in mechanical systems. A new method named a few samples and combined the probability and the fuzzy mathematics has been set up to explore the unknown system of a few samples with the help of the life samples of the known probabilistic distribution. The method reduced the traditional reliability test samples by some auxiliary information. It converted lots of reliability test into a few samples. This paper introduced two engineering examples to verify the method as an effective way to carry out the reliability tests. It provides the theory basis for the reliability of complicated mechanical systems. The loss caused by fatigue failure is as high as 3%~4% of Gross National Product. Fatigue fracture owned to the cycle load accounts for 95% of the total number of mechanical structure failure. Unfortunately no evidence of failure is observed when fatigue failure reaches its life. This causes human casualties and great economic loss. So the reliability research on complicated mechanical systems has to realize from reliability assessment to the active reliability design. One of the key problems of the active reliability design for complicated mechanical system is to reduce the great waste of the resources used in the probability reliability test. For example, dozens of the high precision gears made of certain material may be run out in the fatigue test which would be running day and night for several months. It has been described many methods about system reliability design in the former research [1-3]. But for many high technological products and important mechanical devices, it is impossible to get the test sample to carry out probability analysis. To exert the advantages of the active reliability design and to realize the green reliability design , the only way is to study small samples in the reliability tests and to develop new technique. It is named as a few samples reliability tests method of the complicated mechanical systems.