A study is described to evaluate the Graduate Record Examination (GRE) as an indicator of learning by undergraduate physics majors. Measures were obtained of the overlap of the Graduate Record Examination in physics with the curriculum as it is taught to physics majors in US colleges and universities. The output of the project, for a given subfieldfand setjof institutions, was two numbersRj, f1andRj, f2. These two numbers estimated the extent to which (1) forRj, f1, the curriculum prepares students at institutionsjto take that part of the GRE which covers subfieldfand (2) forRj, f2, the GRE covers the materials in subfieldftaught in the curriculum of institutionsj. The subfieldsf were mechanics, electricity and magnetism, statistical mechanics and thermodynamics, modern physics, and ‘‘other’’ (including experimental, mathematical, and solid‐state physics). The institutionsjwere grouped into four categories: ‘‘top graduate’’ (those with the 20 best‐ranked graduate programs); ‘‘top undergraduate’’ (those participating in the Oberlin Conference on College Science); ‘‘random graduate’’ (other institutions with graduate programs); and ‘‘random undergraduate’’ (other institutions without graduate programs). Generally the results show that the GRE does not cover the entire curriculum (low values ofR2) for any institutional type. With regard to the coverage of the GRE by the curriculum (R1values), there are significant differences between institutions of different types. Institutions with graduate programs have curricula that cover the modern physics and statistical mechanics aspects of the GRE significantly better than institutions without graduate programs.