Making assumptions is an important step in solving many real-world problems. This study investigated whether participants who could solve well-defined physics problems could also solve a real-world physics problem that involved the need to make assumptions. The participants, who all had at least a BA in physics, were videotaped "thinking aloud" while solving three well-defined and one real-world problem and then interviewed about the problem-solving process. All the problems dealt with the same scientific content. The recordings were analyzed to identify similarities and differences in the ways the problems were solved and to see which steps in the solution of the problems posed the greatest cognitive difficulty for each participant. Results indicate that (a) the process of making the constraining assumptions needed to convert the real-world problem into a well-defined one was the most difficult step for all, and (b) only the participants who had prior experience making constraining assumptions were able to make the needed assumptions and solve the real-world problem. These findings suggest a need to support physics students develop this important skill. (Contains 3 tables and 4 figures.)