OBJECTIVES: To review statistical methods and their applications in situations involving multiple causes of failure when studying critical care outcomes (death and utilization of mechanical ventilation [MV]/extubation, hospital and intensive care unit length of stay). METHODS: A targeted literature search was conducted to identify methods that address issues of competing risk of death, censoring, and other statistical considerations applicable to the critical care setting. A total of 31 relevant articles were reviewed. To illustrate the performance of recommended methods, we conducted a simulation study to compare results from standard Cox and competing risk models. Relationships among time-to-event variables and covariates were defined a priori based on the relevant literature. Two standard Cox models were fit for time-to-event analysis (time from MV to cure and time from MV to extubation), which ignore subsequent events, such as death. We also conducted analysis of time-to-event by competing risks, with events of interest either time-to-cure or extubation duration, treating death as a competing risk. RESULTS: Based on our targeted literature search, we concluded that competing risks can be used instead of Cox survival models. Our simulations showed that Cox models appear to overestimate the effects of the treatment variable on the risk of cure by 7% and on the risk of extubation by 2% compared with competing risk estimates. Treating patients who died as if they were censored would lead to overestimation of the hazard rate in the standard Cox models. CONCLUSIONS: Competing events are common in critical care research. In this context, death is considered a competing event, which prevents other events of interest from occurring, and should not be treated as censoring. According to our literature search and simulation study, competing risk models should be used instead of standard Cox regressions in the presence of one or more competing risks.