Subjectivity, which is related to fuzziness, is inherent in system reliability analysis. Therefore, it is necessary to construct a system reliability analysis model into which subjectivity can be incorporated. This paper describes a model of subjective analysis of system reliability. In the model, information about system component reliability and human reliability is expressed by natural language or by numerical values. The information is translated into a fuzzy set that is defined on the unit interval. This fuzzy set is considered as a subjective reliability measure. The model applies a fault tree analysis and uses parametrized and and or operations of the fuzzy sets. These operations can cover the range of the reliability estimate from the most pessimistic estimate through the most optimistic one and reflect analyst's subjectivity towards the analyzed system. The model also considers dependence between subsystems and possibility of subsystem failure leading to a system accident. Analysis results are expressed by the use of natural language expressions. In this model various results can be obtained even if reliability of the same system is analyzed since analyst's subjectivity is incorporated into the analysis. This model considers two kinds of mutual agreements that are an intersection and a union of fuzzy sets, and the weighted average of fuzzy sets.This paper also proposes a prototype of an analyzer based on the present model in order to show the feasibility of the model. The analyzer is implemented on a workstation using the C language and the X-windows system. This analyzer is explained with the help of an example. Subjectivity can be incorporated into the system reliability analysis by the present model, and the subjective reliability analysis can be performed by the use of this analyzer. Finally, this paper compares the present method and the probabilistic method and emphasizes the usefulness of the present consideration.