Five‐year overall survival of stage III colorectal cancer (CRC) patients treated with standard adjuvant chemotherapy (ACHT) is highly variable. Genomic biomarkers and/or transcriptomic profiles identified lack of adequate validation. Aim of our study was to identify and validate molecular biomarkers predictive of ACHT response in stage III CRC patients by a transcriptomic approach. From a series of CRC patients who received ACHT, two stage III extreme cohorts (unfavorable vs. favorable prognosis) were selected. RNA‐sequencing was performed from fresh frozen explants. Tumors were characterized for somatic mutations. Validation was performed in stage III CRC patients extracted from two GEO datasets. According to disease‐free survival (DFS), 108 differentially expressed genes (104/4 up/downregulated in the unfavorable prognosis group) were identified. Among 104 upregulated genes, 42 belonged to olfactory signaling pathways, 62 were classified as pseudogenes (n = 17), uncharacterized noncoding RNA (n = 10), immune response genes (n = 4), microRNA (n = 1), cancer‐related genes (n = 14) and cancer‐unrelated genes (n = 16). Three out of four down‐regulated genes were cancer‐related. Mutational status (i.e., RAS, BRAF, PIK3CA) did not differ among the cohorts. In the validation cohort, multivariate analysis showed high PNN and KCNQ1OT1 expression predictive of shorter DFS in ACHT treated patients (p = 0.018 and p = 0.014, respectively); no difference was observed in untreated patients. This is the first study that identifies by a transcriptomic approach and validates PNN and KCNQ1OT1 as molecular biomarkers predictive of chemotherapy response in stage III CRC patients. After a further validation in an independent cohort, PNN and KCNQ1OT1 evaluation could be proposed to prospectively identify stage III CRC patients benefiting from ACHT.