The hypothesis that epithelial-mesenchymal transition (EMT) might be a contributor to the accumulation of fibroblasts and myofibroblasts (MFs) in the kidney during fibrogenesis was postulated 15 years ago. This paradigm offered an elegant explanation of how the loss of epithelial functions is coupled to the gain of deleterious mesenchymal functions; for example, excessive matrix deposition. Moreover, it interpreted chronic kidney disease in a developmental context: because the tubular epithelium originates from the metanephric mesenchyme, EMT can be viewed as a dedifferentiation process in response to injury, which might serve healing or--if dysregulated--might facilitate fibrosis. Several observations support the role of EMT in renal fibrosis: (1) Tubular cells can transform to fibroblasts and MFs in vitro. (2) Histological 'snapshots' reveal the coexistence of epithelial and mesenchymal markers in transitioning tubular cells in fibrosis models and human kidney diseases. (3) Early lineage-tracing experiments detected mesenchymal markers in the genetically tagged epithelium. However, the paradigm has been recently challenged; new fate-mapping studies found no evidence for the expression of (myo)fibroblast markers in the epithelium during fibrogenesis. This review summarizes the key findings and caveats, aiming at a balanced view, which neither overestimates the role of the epithelium in MF generation nor denies the importance of epithelial plasticity in fibrogenesis.