Measuring response times has been a staple for evaluating masked semantic priming. Its efficacy, however, has been challenged on several grounds — reported effect sizes of these studies are relatively small, and priming effects pertaining to response time measures are difficult to be replicated. Here, we report a complementary method — recording trajectories of a computer cursor. Participants judged whether two digits were the same or different, preceded by a briefly presented masked prime. Each prime had either positive or negative connotations, and the priming effects were evaluated either by response times or cursor trajectories associated with the area under the curve. Results indicate that the effect size of the congruency effect measured by cursor trajectories (i.e. area under the curve) was far greater than that measured by response times, suggesting that the cursor trajectory measure is more sensitive to masked semantic priming than the response time measure.