This study proposes a new model to measure learners’ vocabulary retention, namely the Tracing Measure of Retention (TMR), as opposed to the conventional method, named as the Simple Comparison of Scores (SCS) in this study. The SCS has been mostly based on a fixed group of target words, which does not properly reflect participants’ individual differences. Moreover, the delayed-test scores may possibly include the words that are not subject to retention. This study attempts to justify the newly proposed model as a solution to these issues by applying both models to an authentic learning activity and comparing the two sets of outcomes. Seventy eight university students participated in an in-class reading activity which was designed to boost incidental vocabulary learning, and the vocabulary tests accompanied. The two measurement models yielded different results, which were compared by means of participants’ proficiency levels as a verification tool. The Pearson correlation analysis and ANOVA tests were carried out to check for statistical significance. The results supported the TMR as a more accurate and practical measure of retention, which would serve more insightful research.