Abstract: According to the equivalence theorem of information quantity provided by the practical test of the parts or subsystems and the equivalence test of the system the test data of the parts or subsystems are converted into the equivalence test data of the system.Furthermore,the computation model of the practical test information of the parts or subsystems and the system equivalence conversion model are formulated and presented.In the light of the basic principle of unit reliability assessment,the general models of entropy method synthesis are established for system reliability first approximate confidence limits.These models include system reliability s-Classical non-randomization optimization first approximate limits (lower,upper and two-side limits),and system reliability s-Bayes first approximate limits (lower and upper limits).An example is given and the assessment results are satisfactory in comparison with the traditional methods.