The literature on stimulus equivalence frequently reports the difficulty of obtaining unequivocal data on equivalence relations and their defining properties, specially symmetry and transitivity, with non-human subjects. This fact has  generated a discussion of the role of language in equivalence class formation. The difficulty to obtain equivalence with non-human subjects could be related to the lack of specific history or to procedural failures to eliminate relevant sources of control, like control by stimuli position, rather than due to subjects language limitations. Therefore, it can be relevant to investigate location as a stimulus. The objective of this experiment was to verify if symmetry would emerge after matching-to-sample training with three sample-comparison relations of location. One monkey Ateles paniscus paniscus experimentally naive was used as subject and a matching-to-sample procedure was used to train the conditional relations. The experiment had four phases: 1) Magazine training; 2) Shaping of the key press response using a panel; 3) Training of the conditional relations. Initially, trials had only the sample and the correct comparison stimuli and, after that, 1 and 2 wrong comparisons were introduced; and 4) Symmetry test. The results were negative. In the test, the subject did not reversed the functions of the stimuli, as expected by symmetry. The subject responded in a manner similar lo the training. Other relations did emerge during the test, but these were similar to the trained relations. This also had been reported with humans. It is possible that the baseline discriminations constitute independent sequences of responses instead of a true matching-to-sample. This could explain the non emergence of symmetry. The use of location as stimulus did not lead lo the emergence of symmetry. Further research with the repetition of training and tests with more relations of location to guarantee conditional control, that could eventually lead to the emergence of symmetry, is underway.  A líleratura da área de equivalêncía de estímulos relata freqüentemente a dífículdade de obtençâo de relaçôes de equivalêncía e de suas relaçôes definidoras, especialmente símetría e transitividade, com nâo-humanos. Urna hípótese levantada na literatura da área é que a posíçâo dos estímulos assume papel controlador, difícultando a tarefa para esses sujeitos. O estudo da própria posiçâo como estímulo se torna entáo relevante. O presente experimento objetivou verificar se simetria pode ser obtida a partir de um treino de trés relaçôes entre posíçôes. Um macaco Ateles paniscus paniscus experimentalmente ingênuo foi utilizado como sujeito e um procedímento de pareamento com o modelo foi usado para treino das relacóes. O experimento constou de 4 fases: 1) treino ao bebedouro, 2) modelagem da resposta de pressâo às chaves do painel, 3) treino das relaçôes condicionais, onde inicialmente apenas a estímulo modelo e a comparaçâo correta estavam presentes e depois as comparaçôes erradas eram introduzidas progressivamente, 4) teste de simetria. Os resultados mostraram que simetría nâo foi obtida. No teste, a sujeito respondeu do modo mais semelhante ao treino quanto possível, demonstrando novas relaçôes entre posíçôes muito parecidas com as relaçôes treinadas. A posiçâo como estímulo nâo facilitou a emergencia da simetria. Os dados apontam para a possível necessidade de reformulaçâo da linha de base, com o objetivo de garantir controle condicional, ou treino da simetria, seguido de treinos de novas linhas de base e noves testes até uma possível emergência.