Abstract The landslide victory of the Democratic Progressive Party in Taiwan's 2016 presidential and legislative elections is often interpreted as a persistent swing in attitudes toward cross-strait relations and Taiwan independence. Popular as this interpretation may be, it still runs the risk of mistaking a short-term reaction to a lame duck president's policy performance for a long-term change in attitudes. This study analyzes the evolution of independence–unification (IU) views in the Taiwan population from 1996 to 2016. After reviewing the literature on political generations in Taiwan, I hypothesize that a long-term cohort succession replacing older prounification generations accounted for the evolution toward proindependence views. By pooling six independent face-to-face surveys into repeated cross-sectional data, this study applies a comprehensive multilevel cross-classified random-effect model of age–period–cohort analysis. Markov chain Monte Carlo estimation results confirm that, besides age and period effects, younger cohorts and the 1999 Taiwan-centered high school curriculum have had significant effects on the change in trends in IU views.