Background: As part of the MEDLINE Retagging Project (1994 to 2006) for The Cochrane Collaboration and the US National Library of Medicine (NLM), the US Cochrane Center (USCC) annually searched MEDLINE on PubMed to identify reports from randomized controlled trials (RCTs) not tagged with the RCT publication type (RCT[pt]). Identified RCT reports were submitted to NLM for ‘‘retagging’’ as RCT[pt]. Objectives: To examine 591 RCTs added to PubMed in 2005 and identified as RCT[pt] by the USCC in 2006, which had been processed by NLM but not tagged RCT[pt], to understand potential explanations for the discrepancy. Methods: Two independent reviewers have completed preliminary classification of the first 100 records, and we will report on all 591. We classified records according to journal characteristics, type of report (e.g. secondary analysis, design and rationale, baseline results, diagnostic test accuracy), and NLM Medical Subject Headings (MeSH) indexing. Reports could be classified as more than one type. Results: We classified 3/100 reports as ‘‘not RCT[pt]’’. Almost all records confirmed as RCT[pt] were from specialty journals (92/97) and in English (96/97). Twenty reports appeared to be straightforward RCTs. The most common report types otherwise were secondary analysis of RCT data (31/97); observational analyses from RCT data (13/97); and the rationale, design, pilot or baseline data for an RCT (21/97). Half the records (46/97) were MeSH tagged ‘‘Randomized Controlled Trials’’ indicating that RCTs were indexed as the topic but not the publication type by NLM; of these, design, baseline, and followup reports were tagged most often (see Table 1). Conclusions: Preliminary analysis indicates that nearly all records processed but not tagged RCT[pt] by NLM and identified by Cochrane as RCTs do describe trials. Because the Retagging Project has been suspended, Cochrane searchers should consider incorporating the MeSH ‘‘Randomized Controlled Trials’’ in their strategies for records entered into PubMed in 2006 and later.