The present study aimed at determining whether (1) children were able to interrupt a strategy execution to switch and choose another better strategy, and (2) their ability to switch strategy within-item improved with age. Third, fifth, and seventh graders performed a computational estimation task in which they had to provide the better estimates to two-digit addition problems (e.g., 32 + 54) while using the rounding-down (e.g., 30 + 50) or the rounding-up strategy (e.g., 40 + 60). After having executing the cued strategy (e.g., 30 + 50) during 1,000 ms, participants were given the opportunity to switch to another better strategy (e.g., 40 + 60) or to repeat the same strategy (e.g., 30 + 50). The results showed that children switched strategies within items, and were able to switch more often when the addition problems were cued with the poorer strategy (e.g., 40 + 60 for 32 + 54) than when cued with the better strategy (e.g., 30 + 50). As they grew up, children based their decisions to switch strategies more often on whether the 1,000-ms strategy execution concerned the better strategy or strategy difficulty (i.e., the rounding-up strategy). These findings have important implications to further understand mechanisms underlying within-item strategy switching as well as strategic variations in children.