E-learning has changed the way of traditional education through anytime and anywhere learning experience. With the widespread adoption of Web services technology, the legacy e-learning systems are being exposed as reusable services on the Web. The increasing e-learning services on the Web prompts a research in selecting the most suitable e-learning service for the learner's needs. Quality of Service (QoS) is a decisive factor to discriminate e-learning services offering similar services/resources. In this paper, the authors broadly classify e-learning systems/applications into three groups based on the nature of learning experience and processes involved. Based on the e-learning system's taxonomy, the paper defines a QoS model with six categories of QoS parameters to discriminate e-learning services. The paper proposes an architecture and ranking mechanism to select the most suitable e-learning service for the learner based on his QoS requirements and preferences.