Background: Tumor tissue (T) mutational analysis represents the standard for metastatic colorectal cancer (mCRC); however, circulating tumor DNA (ctDNA) detected by liquid biopsy in plasma (PL) can better represent tumor heterogeneity. Methods: mCRC patients undergoing standard first-line chemotherapy with known T-KRAS/NRAS/BRAF status were enrolled in the present prospective study. PL mutations were assessed within 2 weeks before chemotherapy start with real time PCR and correlated with T status and Progression free survival (PFS). Clinical and biochemical variables including also total number of tumor lesions (TNL) and the sum of maximum diameter (SMD) of all lesions were assessed as potential predictors of T/PL discordance. RESULTS: Among 45 enrolled patients, all BRAF mutations were concordant between T and PL and there were 20% of patients RAS discordant: 9% wild type in T and mutated in PL and 11% mutated in T and wild type in PL. T mutations were significantly associated to median PFS (mPFS of 4.5, 8.3 and 22.9 months for T-BRAF mutated, T-RAS mutated, and T-wild type patients, respectively, p for trend 0.00014). PL mutations further refined prognosis: RAS wild type in T and mutated in PL had significantly shorter PFS than concordant RAS wild type in T and PL: mPFS 9.6 vs. 23.3 months, respectively, p = 0.02. Patients RAS mutated in T and wild type in PL had longer PFS than concordant RAS mutated in T and PL: 24.4 vs. 7.8 months, respectively, p = 0.008. At a multivariate cox regression analysis for PFS, PL mutations were independent prognostic factor superior to T analysis (HR 0.13, p = 0.0008). At multivariate logistic regression analysis TNL and SMD were significant predictors of discordant cases. Conclusions: PL mutational analysis allows a better prognostication than T analysis alone and could help in mCRC treatment management.