Determination of drug safety and tolerability is usually based on the frequency of certain key adverse events (AEs) rather than the frequency of all-grade toxicities. We assessed the reporting of key AEs in oncology randomized, controlled trials (RCTs) and compared that with the expectations of the European Organization for Research and Treatment of Cancer (EORTC) membership.RCTs reports published between 2007 and 2011 were reviewed regarding the reporting of key AEs, namely: grade 3/4 AEs, grade 5 AEs, and AEs resulting in study withdrawal or in dose reduction. Study characteristics associated with better reporting of key AEs were investigated. Finally, a survey was conducted among the EORTC membership to determine their expectations on key AEs reporting.Although the frequency of grade 3/4 was reported in most reports (96%), only 17% of them described the reporting threshold above which grade 3/4 AEs were included for reporting, raising the possibility that important but less frequent grade 3/4 AEs might be underreported. Frequency and nature of grade 5 AEs were adequately reported in 161 (50%) of manuscripts; AEs leading to study withdrawal in 61 manuscripts (19%); and AEs leading to dose reduction in 43 manuscripts (13%). In contrast, most EORTC members expected a comprehensive reporting of grade 5 AEs (96% of EORTC member's responses), AEs leading to study withdrawal (86%) and AEs leading to dose reduction (70%). In multivariate analysis, frequencies of grade 5 AEs were less frequently reported in European trials (P = 0.004). Frequencies of AEs leading to withdrawals were more frequently reported in trials funded by industry (P = 0.005) and in trials including patients with breast or urological cancers (P = 0.006).These findings suggest that current practice of key AEs reporting remains highly variable and sometimes inadequate in oncology RCTs reports. Current standards for safety reporting in RCTs should be revised to emphasize the importance of key AEs reporting.