Computational thinking (CT) is a fundamental skill for students, and assessment is a critical factor in education. However, there is a lack of effective approaches to CT assessment. Therefore, we designed the Three-Dimensional Integrated Assessment (TDIA) framework in this article. The TDIA has two aims: one was to integrate three dimensions (directionality, openness, and process) into the design of effective assessment tasks; and the other was to assess comprehensively the three dimensions of CT including computational concepts, practices, and perspectives. Guided by the TDIA framework, we designed three pairs of tasks: closed forward tasks and closed reverse tasks, semiopen forward tasks and semiopen reverse tasks, and open tasks with a creative design report and open tasks without a creative design report. To further confirm each task's applicability and its advantages and disadvantages, we conducted a test experiment at the end of the autumn semester in 2014 in a primary school for 3 weeks. The results indicated that (a) the reverse tasks were not more superior than the forward tasks; (b) the semi-open tasks and the open tasks were more effective than the closed tasks, and the semi-open tasks had higher difficulty and discrimination than the others; (c) the self-reports provided a helpful function for learning diagnosis and guidance; (d) the scores had no significant difference between the schoolboys and the schoolgirls in all six tasks; and (e) the six tasks' difficulty and discrimination were all acceptable, and the semi-open tasks had higher difficulty and discrimination than the others. To effectively apply them, the following suggestions for teachers to design computational tasks are proposed: motivating students' interest and enthusiasm, incorporating semi-finished artifacts, involving learning diagnosis and guidance, and including multiple types of tasks in an assessment.