Learning performance assessment aims to evaluate what knowledge learners have acquired from teaching activities. Objective technical measures of learning performance are difficult to develop, but are extremely important for both teachers and learners. Learning performance assessment using learning portfolios or web server log data is becoming an essential research issue in web-based learning, owing to the rapid growth of e-learning systems and real application in teaching scenes. The traditional summative evaluation by performing examinations or feedback forms is usually employed to evaluate the learning performance for both the traditional classroom learning and the web-based learning. However, summative evaluation only considers final learning outcomes without considering learning processes of learners. This study presents a learning performance assessment scheme by combining four computational intelligence theories, i.e., the proposed refined K-means algorithm, the neuro-fuzzy classifier, the proposed feature reduction scheme, and fuzzy inference, to identify the learning performance assessment rules using the web-based learning portfolios of an individual learner. Experimental results indicate that the evaluation results of the proposed scheme are very close to those of summative assessment results of grade levels. In other words, this scheme can help teachers to assess individual learners precisely utilizing only the learning portfolios in a web-based learning environment. Additionally, teachers can devote themselves to teaching and designing courseware since they save a lot of time in evaluating learning. This idea can be beneficially applied to immediately examine the learning progress of learners, and to perform interactively control learning for e-learning systems. More significantly, teachers could understand the factors influencing learning performance in a web-based learning environment according to the obtained interpretable learning performance assessment rules. (Contains 10 figures and 14 tables.)