This study aims to examine the effect of using computerized concept maps during the pre-writing phase on learners' writing performance. The research questions were: (1) What are the impacts of different computerized concept mapping treatments (no-mapping, individual-mapping, and cooperative-mapping) on writing performance for learners of different writing proficiencies (high-level, middle-level, and low-level)? (2) Does the quality of the concept maps constructed cooperatively exceed the quality of the concept maps constructed individually? (3) Does the map quality correlate to the learner's writing performance? Ninety-four freshmen enrolled in an English course were divided into high-level, middle-level, and low-level learners according to their baseline writing scores. The experimental part of the study took place over two-hour class periods for nine weeks. All participants went through all three treatments for accomplishing three writing assignments. It was found that both computerized mapping treatments had equally positive effects on low-level and middle-level learners compared with the no-mapping treatment. However, high-level learners performed significantly better with the individual-mapping treatment than with the other two treatments. The quality of the concept maps constructed cooperatively exceeded the quality of those constructed individually. The quality of the maps was also correlated to the learners' writing performance. (Contains 2 figures and 11 tables.)