This paper attempts a fundamental analysis of the nature of research into e-learning and the role that theory plays in this. We examine "research" in broad terms, and the nature of phenomena in general. We identify that e-learning is an artificial phenomenon, and that research approaches need to be cognisant of the design elements in e-learning, and the cyclical nature of e-learning development. We identify various desired research outcomes which are appropriate at each stage of the e-learning lifecycle, and argue that studies of e-learning involve a mixture of evaluation and research. We discuss e-learning evaluation research in the context of different disciplinary and interdisciplinary research approaches, recognising that there is no one "right" way to do e-learning evaluation research. However, we recognise that there is a varying mixture of a "search for fundamental understanding" and "consideration of use" in e-learning evaluation research. We use these considerations to discuss the role of theory in educational research, and, in particular, in e-learning evaluation research, before applying the preceding arguments to the e-learning lifecycle, identifying five different forms of evaluation research. (Contains 3 tables.)