The management of cancer patients in clinical practice is undergoing a revolution due to the introduction of effective immunotherapy. Several different strategies currently rely on harnessing the immune system to fight tumors, leading to unprecedented therapeutic success. However, immunotherapy has also created the need to revisit the assessment of treatment benefit, possibly because its mechanism of action leads to effects on short- and long-term outcomes in a manner that differs from those of other modalities. For example, clinical trials of immune checkpoint inhibitors have disclosed novel issues related to the assessment of objective responses and have reinstated overall survival as a reliable primary endpoint, often to the detriment of progression-free survival (PFS). Moreover, pseudoprogression has become an issue in some tumor types, and the duration of responses has emerged as a relevant endpoint. In parallel, the limitations of current statistical methods of survival analysis that are based on the proportional-hazards assumption have become more apparent, and novel methods have been proposed.