Four experiments tested how well students in a college algebra class could use the solution of a distance, mixture, or work problem to solve other problems in the same category. The solution could either be applied directly to solve equivalent problems or had to be slightly modified to solve similar problems. Students in Experiment 1 could not use the solution to produce more correct solutions on either equivalent or similar problems. Experiments 2 and 3 demonstrated that either allowing students to consult the solution as they worked on the test problems or providing more elaborate solutions improved transfer to equivalent problems but did not improve transfer to similar problems. In Experiment 4 there was some transfer to similar problems that differed in complexity, but students relied too much on a syntactic approach in which they filled in the "slots" of an equation.