Reliability models based on Markov chain (Except in queuing systems) have extensive
applications in electrical and electronic devices. In this paper we consider a system with n parallel and
identical elements with constant failure rates (failure rates are exponentially distributed) and the elements
are non repairable. The failure rates increase when some elements are failed. The system works until at
least k elements work. Since the data for the failure rates are either based on historical data or the judgment
of the experts, therefore uncertainty is inherent in the information. These uncertainties can be expressed as
fuzzy rates. For simplicity we consider a triangular fuzzy number either to quantify a linguistic expression
or if estimated quantitative data is available, this fuzzy numbers can be calculated by point estimation and
% (1-b) confidence interval of the parameters of the failure rates. The system of equations are established
and the exact equations are sought for the parameters like MTTF and the probability that system working at
the time t. A numerical example has been solved to demonstrate the procedure which clarifies the
theoretical development. Use of fuzzy parameters to model the uncertainties in the system causes this
model to tackle more realistic situations.