The Banff classification is regularly updated to integrate recent advances in the characterization of kidney allograft rejection, gathering novel diagnostic, prognostic, and theragnostic data into a diagnostic and pathogenesis-based framework. Despite ongoing research on noninvasive biomarkers of kidney rejection, the Banff classification remains, to date, biopsy-centered, primarily relying on a semiquantitative histological scoring system that overall lacks reproducibility and granularity. Besides, the ability of histopathological injuries and transcriptomics analyses from bulk tissue to accurately infer the pathogenesis of rejection is questioned. This review discusses findings from past, current, and emerging innovative tools that have the potential to enhance the characterization of allograft rejection from tissue samples. First, the digitalization of pathological workflows and the rise of deep learning should yield more reproducible and quantitative results from routine slides. Additionally, novel histomorphometric features of kidney rejection could be discovered with an overall genuine clinical implementation perspective. Second, multiplex immunohistochemistry enables in-depth in situ phenotyping of cells from formalin-fixed samples, which can decipher the heterogeneity of the immune infiltrate during kidney allograft rejection. Third, transcriptomics from bulk tissue is gradually integrated into the Banff classification, and its specific context of use is currently under extensive consideration. Finally, single-cell transcriptomics and spatial transcriptomics from formalin-fixed and paraffin-embedded samples are emerging techniques capable of producing up to genome-wide data with unprecedented precision levels. Combining all these approaches gives us hope for novel advances that will address the current blind spots of the Banff system.< Réduire