This article examines elementary students' perspectives on the engagement potential of particular mathematical problems and students' views on their general classroom problem activities. Third-, fifth-, and seventh-grade children from different reform-oriented classrooms were individually interviewed about (a) how they would improve their classroom problem-solving activities and (b) the problems they find the most and least potentially engaging when presented with a range of routine and nonroutine problems. The children requested more relevant, meaningful, and interesting problem experiences in their classrooms, and the fifth and seventh graders requested more representational materials. The children's criteria for determining potentially engaging andnonengaging problems primarily pertained to problem structure and perceived cognitive demands. The nonroutine examples that focused on important reasoning processes and did not involve computation had the greatest engagement potential, while the computational problems had the least appeal.