We tested three squirrel monkeys [ages 2.56 years, SD=1.0], on their ability to associate visual, arbitrary symbols with foods. These subjects live together in large, planted enclosures and were free to enter and leave the laboratory at will; therefore, trial participation was voluntary. In a food-preference test, five foods were presented in pairs. Food pairings were position-controlled and counterbalanced, totalling ten binary choices each was presented ten times in a pseudo-randomized order (totalling 100 trials). In a symbol condition, small plastic lids were marked with one of five arbitrary symbols, each representing one of the five foods, hiding the corresponding food underneath. Following a conditioning phase, the symbol-lids were presented pseudo-randomly in pairs, like the trials in the food preference condition. The subjects could lift one lid and retrieve the food. Two sets of 100 trials were completed. All Pearson's correlation analyses showed that the three variables (food-preferences and two test sets of symbolpreferences)were strongly correlated for each subject, r(3)>0.90, P<0.05; except for Squizzle's first food/symbol comparison, r(3)=0.83, P>0.05. This research shows that Saimiri, living in an enriched environment, whose general diet is not reduced or specifically managed, can learn multiple arbitrary symbol food-symbols quickly, with relatively low-level exposures to the symbol/referent relationships, and select these in accordance with actual food preferences.