A report from the Board on Research Data and Information of the National Academies' National Research Council, entitled released in November 2012 is available for free download (after registering with an email and password) at http://www.nap.edu/catalog.php?record_id=13564. The 239 page report features papers by a wide range of data management experts.
The International Polar Year Data and Information Service notes that "data citation is a developing practice." At http://www.atlantapremiere.com/ the format based on the Chicago Manual of Style 15th edition is: Authors or group. Date of the release. Title of the data set. Editor or compiler. Place of Publication. Data Publisher. Access date. URI or other distribution method.
"Data publishers (e.g. data centers) have a responsibility to work with data providers and science teams to develop the actual content of the citation."
Examples taken from the IPY page [accessed 2010 Sept 16]
Cavalieri, D., C. Parkinson, P. Gloersen, and H. J. Zwally. 1996, updated 2006. Sea ice concentrations from Nimbus-7 SMMR and DMSP SSM/I passive microwave data, March 2002–Sept. 2003. Boulder, Colorado USA: National Snow and Ice Data Center. Data set accessed 2008-05-14 at http://nsidc.org/data/nsidc-0051.html.
König-Langlo, Gert and Hatwig Gernandt. 2006. Compilation of radiosonde data from the Antarctic Georg-Forster station of the German Democratic Republic from 1985 to 1992. Bremerhaven, Germany: Alfred Wegener Institute for Polar and Marine Research Data set accessed 2008-05-22. doi:10.1594/PANGAEA.547983
The Digital Curation Centre in the UK has produced a large body of guidelines and advice for data management. This latest report by Alexander Ball and Monica Duke focuses on best practices in data citation, citing data at the most useful level.
Ball, A. & Duke, M. (2011). How to Cite Datasets and Link to Publications. DCC How-to Guides. Edinburgh: Digital Curation Centre. Available online: http://www.dcc.ac.uk/resources/how-guides
The OECD recently (rev. 2010 Feb.) published a white paper about making their datasets citeable. Noting that much of their data are cited without granularity so that the reader cannot easily find the data upon which the analyses and inferences have been made. See http://dx.doi.org/10.1787/603233448430 for the full report.
DataCite, www.datacite.org, is an international collaboration of university libraries and information centers that is supporting making datasets discoverable and citeable. Digital Object Identifiers (DOI) are assigned to data by DOI Registration Agencies. DataCite and CrossRef are two DOI registrants.