A software component graph, where a node represents a component and an edge represents a use-relation between components, is widely used for analysis methods of software engineering. It is said that a graph is characterized by its degree distribution. In this paper, we investigate software component graphs composed of Java classes, to seek whether the degree distribution follows so-called the power-law, which is a fundamental characteristic of various kinds of graphs in different fields. We found that the in-degree distribution follows the power-law and the out-degree distribution does not follow the power-law. In a software component graph with about 180 thousand components, just a few of the components have more than ten thousand in-degrees while most of the components have only one or zero in-degree.