We report on a concept inventory for special relativity: the development process, data analysis methods, and results from an introductory relativity class. The Relativity Concept Inventory tests understanding of relativistic concepts. An unusual feature is confidence testing for each question. This can provide additional information; for example, high confidence correlated with incorrect answers suggests a misconception. A novel aspect of our data analysis is the use of Monte Carlo simulations to determine the significance of correlations. This approach is particularly useful for small sample sizes, such as ours. Our results show a gender bias that was not present in course assessment, similar to that reported for the Force Concept Inventory. (Contains 5 tables and 2 figures.)