Background: The US satellite of the Cochrane Eyes and Vision Group (CEVG@US) received funding from the National Eye Institute, NIH, to handsearch the US-based vision science literature for randomized controlled trials (RCTs) and controlled clinical trials (CCTs). Objectives: To compare RCTs and CCTs identified by handsearching with articles assigned MEDLINE publication type [randomized controlled trials (PT-RCT)] and [controlled clinical trials (PT-CCT)). Methods: We identified RCTs and CCTs by handsearch and subsequently searched MEDLINE for those published 1966 and later. We determined the proportion of articles identified by handsearching that were classified in MEDLINE as PT-RCT or PT-CCT, and the proportion that were indexed as PT-RCT or PT-CCT, but not classified by us as RCT or CCT. Results: We identified 2,215 RCTS and 469 CCTS by handsearching 223 journal-years, excluding trials published before 1966 and in publications not indexed in MEDLINE. Most reports classified by us as RCTs had been indexed in MEDLINE as PT-RCT (73.9%; 1,637/2,215), but most CCTs were not indexed as PT-CCT (21.3%; 100/469). Conversely, many articles were indexed as PT-RCT (n = 161) or PT-CCT (n = 115) although we did not classify them as trials. Sometimes the reports we classified as describing RCTs were indexed as PT-CCT (n = 22) and sometimes CCTs were classified as PT-RCT (n = 24). In a subset of articles from one journal, more articles were incorrectly indexed with PT-RCT or PT-CCT when published in 2000 or after compared with those dated before 2000 (40 versus 10). Conclusions: A meaningful proportion of articles we identified and classified as RCT and CCT are not indexed by these publication types in MEDLINE. Handsearching is a necessary adjunct to electronic database searching for trials for systematic reviews.