There are many sampling-based motion planning methods that model the connectivity of a robot's configuration space (C-space) with a graph (or roadmap) whose nodes are valid configurations and whose edges represent sequences of valid configurations for moving between nodes. Many planning strategies have been proposed to construct roadmaps. Although their authors have discussed planner strengths and have presented experimental evaluations, the emphasis is usually put on performance traits. There is a need for metrics to strengthen qualitative analysis of planners. Roadmap comparison can be applied to study the evolution of a roadmap as it is built and to compare roadmaps made with different planners. We propose several metrics to compare roadmaps as a basis for analyzing the roadmap generation process and for comparing the relative strengths and weaknesses of different planning strategies.