The growing expansion of immuno-oncology has led to identify and implement the viability of predictive and on-treatment biomarkers to improve cancer patient management. Despite the encouraging great clinical benefit obtained by immunotherapy sustained by accurate procedure to enroll, there are many patients who do not respond or respond poorly developing a hyper-progressive disease with a worse prognosis. Different categories of biomarkers, predictive of response, and on-treatment have been gradually explored to elucidate resistance mechanisms and guide therapy decisions. The identification of reliable biomarkers is limited by our incomplete understanding both of complex immune response to cancer, as well as of immune-editing contribution to clonal changes occurring within tumor microenvironment. To ensure the high reproducibility of the results, it is necessary develop a multi-marker model or workflow with comprehensive immune tumor profiling of the predictive signatures and prospective markers for escalation changes upon therapy with a counterpart in blood. Besides their proven clinical relevance, candidate biomarkers should fulfill strict specificity and sensitivity criteria; also they must be reproducible, easy to perform and to interpret, and cost-effective. Despite the high number of promising biomarkers, their spread in the clinical routine is still very slow.