AbstractThe current grading systems for computer programming assignments have taken correctness, efficiency, complexity, and maintainability into account. Of these four components, the most important measurement is the correctness. However, the existing grading systems still have some drawbacks. It is hard to measure college students' overall programming skills based only on their answers to a single form of programming questions. The author proposes a new approach by presenting multiple forms of computer programming questions, such as statement filling‐in, program modifying and algorithm designing, and by providing an automated grading algorithm to measure the correctness, time efficiency, space efficiency, complexity, and robustness. This proposed automated grading method has been employed successfully in the development of C programming and Delphi programming exam systems for Computer Application Ability Exam for College Students in Fujian. It has also been applied to the development of an online programming self‐testing system for Data Structure course which is offered by Fujian Agriculture and Forestry University. © 2010 Wiley Periodicals, Inc. Comput Appl Eng Educ 21: 484–490, 2013