Computer-based assessment allows for the monitoring of test-takers’ behaviors. Previous research has made efforts to identify their behavioral patterns or investigate potentially influential factors. However, very few studies have examined the changes in behavioral patterns when completing a sequence of tasks. To fill this research gap, we aim to 1) explore different disengaged and engaged behavioral patterns in completing problem-solving tasks, 2) investigate changing behavioral patterns across different tasks, and 3) examine both task and individual characteristics relating to the changing behavioral patterns on all tasks. Data used in this study were drawn from log data of the U.S. Piaac 2012 (N = 660). We first identified two disengaged behaviors (i.e., rapid omit and rapid perfunctory action). Results of cluster analysis also show six types of engaged behaviors (i.e., proficient, exploring, hasty, reflecting, struggling, and passive). Overall, the transition of behavior patterns of adults in the first 8 tasks was relatively stable, while it was more complex in the following 6 tasks. The results of multinomial regression analysis indicate that adults' behaviors in completing problem-solving in technology-rich environments are mostly due to task-related factors, and individual-related factors including age, education attainment, and gender, as well as learning abilities and the use of ICT skills. The findings from this study provide several implications for future vocational training in the context of ICT needs.