Although concept maps have been used as alternative assessment methods in education, there has been an ongoing debate on how to evaluate students' concept maps. This study discusses how to evaluate students' concept maps as an assessment tool before and after 15 computer-based Predict-Observe-Explain (CB-POE) tasks related to acid-base chemistry. 12 high school students participated in this study. Students' concept maps were evaluated taking into account three parts: qualitative, quantitative and representational level. The results gathered from the quantitative analysis of the pre and post concept maps were analyzed using SPSS 17.0. The results showed that there is a significant difference between students' pre and post concept map scores (z = 3.05; p < 0.05). From the qualitative analysis of the concept maps, the majority of the students constructed their pre and post concept maps non-hierarchically; while they drew more interconnected concept maps after the CB-POE tasks. Regarding their representational level, the students used the macroscopic level more often than microscopic and symbolic levels in both their pre and post concept maps. Nonetheless, they increased the number of macroscopic, microscopic and symbolic level representations in their post concept maps. The paper concludes with suggestions and implications for educators and researchers to improve the quality of concept map evaluation.