A quasi-experiment was conducted at an elementary school to investigate if guided collaboration would facilitate programming learning of 6 graders. Sixty-six students of two intact classes learned to program in KPL (kid's programming language) for 18 weeks during the experiment. One class was randomly assigned to the control group (i.e., free-collaboration) and the other to the experimental group (i.e., guided-collaboration). Students in both groups formed heterogeneous teams of 3 persons. The guided-collaboration teams in experimental group were provided with a worksheet for every programming task. The worksheet contained a set of task-specific guiding questions to guide students through the problem-solving process in a systematic and disciplined manner. An analysis of test scores showed that the experimental group significantly outperformed the control group in the achievement tests, suggesting that the guiding questions were useful in enhancing students' comprehension of programming concepts and developing their programming skills. A comparison of the amount of time spent at each problem-solving step showed that the experimental group spent significantly more time than the control group at the analysis/design/coding step as well as the reflection step. Conversely, the control group spent considerably more time at the debugging step. With the help of the guiding questions, students in the experimental group were also more able to conduct meaningful discussions. (Contains 6 tables and 2 figures.)