We investigated the effects of two context variables, presentation format (icon arrays or numerical frequencies) and time limitation (limited or unlimited time), on the proportional reasoning abilities of children aged 7 and 10 years, as well as adults. Participants had to select, between two sets of tokens, the one that offered the highest likelihood of drawing a gold token, that is, the set of elements with the greater proportion of gold tokens. Results show that participants performed better in the unlimited time condition. Moreover, besides a general developmental improvement in accuracy, our results show that younger children performed better when proportions were presented as icon arrays, whereas older children and adults were similarly accurate in the two presentation format conditions.