Actual assessment methods for identifying adverse events (AEs) in clinical trials have received less scrutiny than underreporting of AEs.To investigate whether AE assessment has changed over time for three psychotropic drugs with turbulent histories of safety concerns since their U.S. approval.From industry-funded published trials of atomoxetine, duloxetine, and olanzapine retrieved from PubMed for 1996-2004 (n = 33) and 2009-2014 (n = 40), verbatim AE assessment and numbers of words describing efficacy and safety assessment were extracted.Closest to drug approval (1996-2004), 77.8% of atomoxetine trials used open-ended questioning only, 50% of duloxetine trials used spontaneous self-report or clinician observation only, and 66.7% of olanzapine trials used a scale (primarily for extrapyramidal symptoms) and one former method. Recent studies (2009-2014) showed less rigor and transparency: 35.3% of atomoxetine and 64.7% of duloxetine studies reported no AE assessment method and 50% of olanzapine studies no longer used scales. Overall, the mean number of words describing efficacy assessment increased from 202 to 309 but decreased from 83 to 63 for safety.Trial methodology for assessing psychotropic drug safety remains an underdeveloped area with major public health implications.