This paper presents reliability characteristics of a complex system using Classical and Bayesian approaches. The system comprises three non-identical units-A, B, C. The first unit-A is arranged in series network with other two units-B & C while the units-B &C are themselves arranged in parallel network. Initially all the units are operative. If during the repair of one unit, other unit also fails then the whole system is put under emergency repair. All failure & repair time distributions are taken as general. Using regenerative point technique, various reliability characteristics are obtained. A particular case has also been discussed where all the failure & repair distributions of times have been taken as Weibull. Classical and Bayesian analysis have also been made for a particular case. Under Classical analysis Maximum Likelihood Estimators (MLE) of the parameters representing the reliability characteristics are obtained. A Bayesian approach is also adopted to evaluate the reliability characteristics. Monte Carlo simulation study is performed to judge the performances of the ML and Bayes Estimator.