Present research in problem solving appears to be primarily concerned with problem-solving methods and with degree of knowledge acquisition. A brief argument is advanced that this conceptualization is incomplete because of failure to consider individual differences among problem solvers (other than in problem-solving methods and extent of knowledge). A viable theory of problem-solving instruction must take into account all three areas. Evidence for the argument is presented in the form of data on problem-solving success in junior high school students with extreme scores on Witkin's field independence-field dependence measure of cognitive style. Problem-solving protocols are examined as a second source of data. Field independent students significantly out-performed field dependent students on the problems. Examination of protocols revealed consistent performance patterns favoring field independent students.