Lung cancer has recently been discovered to be an immunological targetable disease, on the basis of the exciting results of the randomized trials with immune checkpoint inhibitors. Nevertheless, the survival benefit appears to not be entirely captured by the usual outcome measures, thus requiring a deep reflection about the appropriateness of the traditional statistical methodologies in this context. The intrinsic biological differences existing both in terms of mechanism of action and kinetic between immunotherapy and chemotherapy or targeted therapy, impact on patients' outcome, requiring a global revolution in the way to design clinical studies with the ideal aim to evolve towards trials carefully 'customized' on the basis of the investigational drug, the specific disease and the biological background. The exciting data recently obtained with immune checkpoint inhibitors, offer an ideal context and background to explore the major questions and future perspectives about the development of immunotherapeutic agents. In this regard, the choice of adequate endpoints, the use of modified statistical methods and the potential introduction of predictive biomarkers for immunotherapy clinical trials, will be discuss in this review in order to provide practical and rationale suggestions aimed to improve the existing model for cancer immunotherapy investigation.