An electronic portfolio system, designed to serve as a resource-based learning space, was tested in a fifth-grade science class. The control-group students accessed a traditional folder-based information display in the system and the experimental-group students accessed a concept map-based information display to review a science portfolio. The student-constructed science portfolio was a result of a collection of digital artefacts such as graphic images, instructional videos and textual files on terms and definitions relevant to the Earth's atmosphere. In the information-processing performance test, the experimental-group students scored significantly higher and spent much less time in finding answers to the questions presented. Also, in the 3-day delayed memory retention tests, the experimental-group students achieved significantly higher scores. The multiple regression models confirm that the visualisation variation is the most significant predictor for student performance (ie, comprehension and access time). Also, the regression models suggest that language-arts proficiency may reduce overall system access time; while higher math proficiency may predict retention scores.