In this paper, we have attempted two ambitious tasks. We have undertaken a wide-ranging survey of the Network learning (NL) literature, and tried to identify the emerging themes of this work. We have selected three of these themes, and in each case tried to identify the main theoretical perspectives in use, the main directions of the studies, and the key ideas being addressed and researched. We have also tried to indicate where the main research effort might be directed in order to help to ‘fill in the gaps’ and achieve some coherence for the theme. Our second major task has arisen from our assertion that the field of Networked Learning research is theoretically fragmented. We have argued that this situation arises because Networked Learning research is a new field, and is drawing upon a wide range of theoretical perspectives. However, unless we can achieve some synthesis of these perspectives we may find it difficult to establish a coherent research programme in the field. We argue that one way of developing some coherence is to make theory and praxis interact explicitly, in other words, to ‘converse’ with each other in our research. By this we mean, to use theory to interrogate praxis, and use praxis to modify and develop theory, thus moving towards perspectives that are changing theory, modifying and improving it. As part of this argument, we have briefly surveyed the current level of Theory–Praxis Conversation, either explicit, or implicit, in the thematic research we have described. It is clear that some outstanding work is being done to make theory work, and to modify it in the light of research into praxis. However, it is also the case that much current Networked Learning research does not interrogate the theory that it uses to contextualise it. We see Theory–Praxis Conversation as a way of thinking explicitly about how we might make the work of interrogating theory in our research more explicit and systematic. In this way, our ‘Quest for Coherence’ may, we hope, help Networked Learning research to climb up to the higher ground, and give us a wider ranging view of learning in networked environments.