The use of stones to crack open encapsulated fruits is one of the most complex forms of tool use in primates, at the cognitive and manipulative levels. The choice of an adequate tool is a critical aspect in cracking behavior. So far, this question has not been experimentally studied in capuchin monkeys out of captivity settings. The present study was conducted to determine which factors affect the choice of stone tools for nut cracking by members of a semi-free ranging group of capuchin monkeys ( Cebus apella) in Tietê Ecological Park, São Paulo, in which the spontaneous and traditional use of stones as tools has being studied for one decade. In this experiment five artificial hammers (stones used to pound the nut) were used, all made of the same material and format, but ranging in weigh between 300g and 1800g. The hammers were placed in a random sequence between two bigger flat stones, used as anvils (hard surfaces serving as support). Nuts of Syagrus romanzoffiana were available ad libitum. The results show that, considering all age classes, there was no preference for hammers based on its position, but there was based on the weight of hammers: the hammer weighting 1300g was used significantly more. Comparisons between age classes revealed that the young individuals, besides the still significant preference for the 1300g hammer, had also a significant preference for hammers in the positions next to the anvils. The adults/subadults had no preference for position, and chose significantly more the two heavier hammers. The influence of a less essential variable for cracking performance (easiness in reaching the tool) in the hammer choice by youngsters can be associated to a less clear discrimination of the critical properties of the tools (the hammer weight) for the right resolution of the task.