This paper offers an investigation into the general message structure and politeness strategies used in email communications between (now convicted) paedophiles in discussions of highly illegal practices and intent. In these emails, the interactants reveal telling characteristics of their ‘community of practice’. Consistent with the view from cognitive anthropology (Wenger, Communities of practice: Learning, meaning, and identity, Cambridge University Press, 1998, IVEY Business Journal, 2004) and sociolinguistic research (Eckert, Linguistic variation as social practice, Blackwell, 2000; Newell-Jones, Whose reality counts? Interprofessional learning through the eyes of Participatory Rural Appraisal, Academy of Higher Education, LTSN. Academy of Higher Education, Health Sciences and Practice, 2005), the members of this speech community signal their ‘in group’ membership and values through a number of linguistic strategies, such as lexical and topic choices. Through these choices, members can not only quickly detect non-members, but can focus on those factors that are central to their community (in this case, avoiding features of talk, such as banter, sarcasm or humour, that are risky in new, internet relationships but would mark more solid relationships). The body of email data used for this research is a corpus taken from a recent paedophile case in the UK. Despite the relatively small dataset, many emails are captured in discourse ‘strings’ (i. e., multiple messages between users, embedded within single emails) and thus capture the consecutive to-and-fro of discourse between these discussants. As such we have clear evidence of the progress between contributions and it is clear how quickly members of this speech community discuss sensitive information, and how quickly they seek to ‘meet up’, despite knowing very little about each other. These data are also packed with descriptive information about the wants, deeds and desires of the email senders, making this a particularly rich corpus. We argue that facework and politeness ought to be considered in the light of the overall target social objectives, rather than how we, or members of other social groups, might perceive such discourse contributions.