We examined the claim that equivalence classes contain all positive elements in a reinforcement contingency by asking whether negative stimuli in a reinforcement contingency will also form an equivalence class, based on their shared function as S-minus stimuli. In Experiment 1, 5 subjects were tested for equivalence for positive and negative stimuli. Testing of positive stimuli preceded testing of negative stimuli. Two of five subjects demonstrated equivalence for positive stimuli, and three subjects demonstrated equivalence for negative stimuli. In Experiment 2, order of testing was reversed. Four of six subjects demonstrated equivalence for positive stimuli, and none demonstrated equivalence for negative stimuli. In Experiment 3, positive and negative stimuli were tested together. Only one of five subject demonstrated equivalence for positive and negative stimuli. These data suggest that negative stimuli may enter an equivalence class, and so Sidman paradigm should be expanded. Order of testing was found as a meaningful variable.