There is now widespread recognition that data are a valuable long-term resource and that making them publicly available is a way to realise their potential value - both as part of the scholarly record or for re-use by others. The Research Information Network (RIN) report, To share or not to share: Publication and quality assurance of research data outputs (June 2008), investigates whether or not researchers make their research data available to others and the issues they encounter when doing so. Importantly, it seeks to do this by seeking the perspectives of researchers themselves. This paper reflects on how this relates to the more top-down literature on the subject. The discussion of the significance of the RIN's main findings is correlated to the four themes of the RIN report. Firstly, it discusses some distinctions in the types of data that should be shared and preserved and what needs to done to do so effectively. Secondly, it reflects on the motivations for and constraints on researchers publishing their data, and how funders and publishers can address them. Thirdly, it reviews some issues around how data are discovered, accessed and re-used. Finally, it discusses the scholarly and technical quality of published data.