The manifold growth model that unifies existing software reliability growth models can cover a wide range of accumulated fault data including various types of data which are difficult to treat with existing models. However, there are problems. For example, it sometimes takes a long time to solve transcendental equations to estimate parameters of the model, and the solution is difficult to obtain in some cases. This paper shows that the most important parameters of the differential equation that defines the manifold growth model can be easily estimated for the given data by using a "Z-graph" to express the Z-equation derived from the differential equation in a 2-dimensional graph. This paper also shows that the most appropriate type of software reliability growth model, or the most appropriate shape of curve, can be determined simply by observing a part of the data sequence from Z-graph, since Z-graph makes it possible that the variation of the value of parameters at any time can be recognized visually. Finally, the effectiveness of the Z-graph is shown by applying the Z-graph to both ideal and actual data.