Previous studies have shown that after being trained on A-B and A-C match-to-sample tasks, adults match not only same-class B and C stimuli (equivalence) but also BC compounds with same-class elements and with different-class elements (BC-BC). The assumption was that the BC-BC performances are based on matching equivalence and nonequivalence relations (equivalence-equivalence). The present study examined if the BC-BC performances could have resulted from matching stimuli with same discriminative (S+, S-) properties. In Experiment 1, the subjects were first trained on a simple Q discrimination task (Q1+/Q2-). After testing the discriminative functions of these stimuli independent from one another, conditional A-B and A-C discriminations were trained. Finally, 3 tests were presented. One test assessed equivalence (B-C, C-B). The second test permitted matching on the basis of same functional (equivalence, nonequivalence) relations (BC-BC). The third test measured the S+ function of compounds with same-class elements and the S-function of compounds with different-class elements (BC/BC), or matching on the basis of same discriminative functions (Q-BC). The order of the tests varied across conditions except that the equivalence test was always presented first. Experiment 2 was the same except that the equivalence was tested last. In both experiments, almost all subjects demonstrated equivalence and selected compounds with same-class elements during a simple BC discrimination test (e.g., B1C1+/B1C2-). Most of these subjects also matched Q and BC stimuli of same discriminative functions (e.g., Q1-B1C1, Q2-B1C2), and BC compounds with same- and with different-class elements (B1C1-B3C3, B2C3-B3C2), more so in Experiment 1 than in Experiment 2. These findings indicate that the BC-BC performances result from matching same discriminative functions. The implications of these findings for equivalence-equivalence as a model for classical analogies are discussed. (Contains 2 figures and 7 tables .)