OBJECTIVES: In the clinical setting, the nocebo phenomenon is where clinical worsening or adverse events occur as a response to a treatment, in a situation in which conditioning from previous treatment exposure and/or expectations of sickness or symptoms lead to sickness and symptoms in a conditioned or expectant individual. The nocebo response may thus be a confounder in clinical treatment and clinical research. There is a need to know how to predict if an individual is likely to be a nocebo responder, and how significant and commonplace the nocebo effect might be. METHODS: An analysis was conducted on nine placebo-controlled, randomized clinical trials of olanzapine for the treatment of bipolar disorder using data from placebo-treated study participants only. Data were analysed to identify participant or study characteristics associated with a nocebo event, defined as any treatment-emergent adverse event (TEAE) or an increase in score from baseline to endpoint for primary measures of clinical symptoms. RESULTS: A total of 1185 participants were randomized to placebo, of whom 806 (68%) reported a TEAE. Hamilton Depression Rating Scale (HDRS) data were only available for 649 placebo-treated participants, of whom 321 (49.5%) demonstrated worsening. Nocebo events were significantly associated with: not being treatment-naïve, younger age, being located in the USA, being a participant in an earlier study, and being classified as obese compared with normal weight. CONCLUSIONS: A pattern to identify nocebo responders did not emerge, although some prognostic variables were associated with a greater probability of nocebo response. There was some evidence to support the role of expectancy as a cause of nocebo reactions.