Assessments were made of the impact of various overt response modes on the problem-solving process. The Luchins Water Jar Problems were presented to undergraduate university students via interactive cathode ray tube computer terminals under four response mode conditions: Mnemonic Code, Multiple Choice (MC), Numerical List, and Constructed Response (CR). Initially, the List and CR modes were least susceptible to mechanization. Overall, the MC mode produced the least mechanized and most efficient performance. The CR mode group spent the greatest time in problem solving without exhibiting superior performance. Analysis of item by mode interactions revealed that across modes the problems were not of equal difficulty. Implications of response modes in problem solving and the design of instructional materials are discussed. (Author/CMV)