This paper reports findings from research that set out to explore virtual action learning (VAL) as an emerging variety of action learning (AL). In bringing together geographically dispersed individuals within and across organizations, and possibly across time, VAL has obvious potential in both educational and commercial contexts. Whilst there is an extensive and growing literature on face-to-face (f2f) AL, at the start of the inquiry (October 2006) there was little evidence of the virtual variety. At the same time, there was considerable interest expressed by educationalists and practitioners in adapting AL to VAL, but a lack of understanding of how to go about it, which appeared to be a barrier to uptake. The research comprised a literature review, a network inquiry and interviews with VAL practitioners. The findings reveal more practice than was anticipated and that VAL is a distinct variety of AL, characterised by its virtual, non-f2f nature, and that it has its own strengths and weaknesses. After discussing the prospects for a virtual form of AL, this paper offers a 6-form classification of VAL, considers some of the theoretical questions that might usefully be pursued in connection with this emerging practice and explores its potential in the light of emerging technologies. The paper concludes that, like AL, VAL is not singular but takes a variety of forms, each with distinct characteristics. The question of whether it is better than, or second best to, f2f AL is debated and we conclude that it should not necessarily be evaluated against f2f AL, but seen as a practice in its own right and located in the broader practices of virtual and network learning and their correspondence to social, cultural, technical and economic change in the wider society. (Contains 2 tables.)