<p>This study examined how priming with positive and negative emotion‑label words (ELWs; e.g., “happy”, “sad”) and emotion‑laden words (ELdWs; e.g., “money”, “death”) influences estimation performance in a two‑digit multiplication task. In Experiment 1, participants applied an assigned strategy, whereas in Experiment 2 they freely chose between strategies. The results revealed several key patterns. First, positive priming (particularly through ELdWs) increased adoption of the down-up strategy (DU strategy, e.g., estimating 43 × 78 as 40 × 80 = 3200), while no clear strategy preference emerged under negative ELWs and ELdWs priming conditions. Second, in Experiment 1, the accuracy (ACC) was lower for positive ELWs than for positive ELdWs priming; moreover, the DU strategy was executed more accurately under positive priming, whereas ACC for the up-down strategy (UD strategy, e.g., estimating 43 × 78 as 50 × 70 = 3500) did not vary by valence. In Experiment 2, ACC showed no significant effects. Reaction time (RT) analyses showed a consistent dissociation across experiments: positive priming sped DU strategy utilization, and negative priming accelerated UD strategy utilization. Furthermore, a significant interaction indicated that the DU strategy was executed faster following ELdW primes, whereas the UD strategy was executed faster following ELW primes. Finally, math anxiety had no significant effect on ACC, RT, or strategy selection adaptivity. Taken together, these findings demonstrate that ELWs and ELdWs priming exert distinct effects on estimation performance, whereas math anxiety exerts no significant influence. This study underscores the importance of distinguishing between ELWs and ELdWs when investigating how emotional language processing modulates cognitive performance.</p>