In basket design trials, a specific challenge is to monitor safety across a range of tumor types that may respond very differently to the investigational treatment. Hence a serious treatment toxicity may be deemed acceptable in tumor types where the treatment efficacy is high enough to warrant its use. In other words, the Independent Data Monitoring Committee (IDMC) should review safety as well as efficacy data to assess the benefit/risk of investigational treatments. Basket design trials can and should generally be randomized, using the standard of care for each tumor type as a control arm (Saad et al., Nature Rev Clin Oncol 14: 317-23, 2017). I will present a novel statistical method for benefit/risk assessment in randomized trials. The method prioritizes several outcomes, and performs generalized pairwise comparisons of patients in the experimental treatment arm with patients in the control arm (Buyse M. Prioritized multiple outcomes. In: Wiley StatsRef: Statistics Reference Online, https://onlinelibrary.wiley.com/doi/10.1002/9781118445112.stat08158, 2018). The method allows several orders of priority to be considered, for instance an efficacy outcome first and a safety outcome second, or the reverse order. In fact, the method allows any number of outcomes to be considered, thus making it possible to present the IDMC with multiple analyses to assess the net benefit of the experimental treatment overall as well as in specific tumor types, taking into account all relevant outcomes simultaneously. This approach is more informative and more relevant than separate analyses that ignore the correlation between the various outcomes considered.