Students' attitudes and approaches to problem solving in physics can profoundly influence their motivation to learn and development of expertise. We developed and validated an Attitudes and Approaches to Problem Solving survey by expanding the Attitudes toward Problem Solving survey of Marx and Cummings and administered it to physics graduate students. Comparison of their responses to the survey questions about problem solving in their own graduate-level courses vs problem solving in the introductory physics courses provides insight into their expertise in introductory and graduate-level physics. The physics graduate students' responses to the survey questions were also compared with those of introductory physics and astronomy students and physics faculty. We find that, even for problem solving in introductory physics, graduate students' responses to some survey questions are less expertlike than those of the physics faculty. Comparison of survey responses of graduate students and introductory students for problem solving in introductory physics suggests that graduate students' responses are in general more expertlike than those of introductory students. However, survey responses suggest that graduate-level problem solving by graduate students on several measures has remarkably similar trends to introductory-level problem solving by introductory students. (Contains 11 figures and 2 tables.)