Attribution of adverse events (AEs) is critical to oncology drug development and the regulatory process. However, processes for determining the causality of AEs are often sub-optimal, unreliable, and inefficient. Thus, we conducted a toxicity-attribution workshop in Silver Springs MD to develop guidance for improving attribution of AEs in oncology clinical trials. Attribution stakeholder experts from regulatory agencies, sponsors and contract research organizations, clinical trial principal investigators, pre-clinical translational scientists, and research staff involved in capturing attribution information participated. We also included patients treated in oncology clinical trials and academic researchers with expertise in attribution. We identified numerous challenges with AE attribution, including the non-informative nature of and burdens associated with the 5-tier system of attribution, increased complexity of trial logistics, costs and time associated with AE attribution data collection, lack of training in attribution for early-career investigators, insufficient baseline assessments, and lack of consistency in the reporting of treatment-related and treatment-emergent AEs in publications and clinical scientific reports. We developed recommendations to improve attribution: we propose transitioning from the present 5-tier system to a 2-3 tier system for attribution, more complete baseline information on patients' clinical status at trial entry, and mechanisms for more rapid sharing of AE information during trials. Oncology societies should develop recommendations and training in attribution of toxicities. We call for further harmonization and synchronization of recommendations regarding causality safety reporting between FDA, EMA and other regulatory agencies. Finally, we suggest that journals maintain or develop standardized requirements for reporting attribution in oncology clinical trials.