Two experiments were conducted to study stimulus equivalence as a function of class size and number of classes. In the first experiment, equivalence was tested in 50 normal adult subjects following a linear series training structure. Subjects were successively assigned to either of 10 groups, exposed to a specific stimulus material. For subjects in which number of classes increased, up to six, B-stimuli served as pictures, while A-, and Cstimuli were Greek letters, and "equivalence" was tested in CA tests. The A-, B-, and C-stimuli were the same for all subjects in whom class members increased up to six, where D-, E-and Fstimuli were Greek letters. FOllowing AB, BC, CD, DE, and FE training , FA, EA, FB, FC, EB, DA, FD, EC, DB, and CA "equivalence" tests were run. In the second experiment, a manyto-one training structure was used to study equivalence as a function of increasing class size without increase number of nodes. The results indicate that the probability of equivalence decreased more as function the number of nodes than as a function of number of classes. Reaction times, particularly to the comparison stimuli, generally increased initially during tests, possibly indicating precurrent problem solving behavior prior to the response to a comparison stimulus.