Background: In order to identify targetable mutations, the Comprehensive Genomic Profiling (CGP) is an important tool. Technologies like extensive genomic NGS panels analyze multiple genes and biomarkers in a single assay. With multi-omics analysis, clinicians could increase the possibility to provide a tailored therapy. In literature, evidence about clinical applicability and performance of NGS are mainly derived from clinical trials and real-world experiences are lacking. Methods: On these bases, we performed a retrospective analysis of CGP data from a cohort comprising 1450 patients (pts) with a diagnosis of gastrointestinal tumor. The primary objective was to estimate adequacy and quality of CGP. As secondary objective, we identified mutations with an available but not already standard of care targeted treatment (TT), defined as gain, and we collected clinical data in those pts. The genomic analysis was performed using the FoundationOne CDx panel. Clinical impact of treated vs not-treated pts with TT was evaluated using 1:1 propensity score matching, according to tumor type, treatment line and gain gene mutation. Survival outcomes were estimated with Kaplan-Meier method and compared with Hazard Ratio (HR), univariate Cox regression. Results: In our population, 1310/1450 samples (90.3%) were adequate for NGS analysis. Failure rates were higher in biopsy samples compared to resections (p < 0.0001); the first cause was insufficient tumor DNA (p = 0.0020) especially for hepato-biliary-pancreatic neoplasms. Low-quality reports observed in 252/1310 (19.2%) were mainly due to inadequate DNA quality or extraction, regardless samples type. Gain molecular alterations were identified in 355/1450 pts (24.5%), corresponding to 28.1% of cases with informative reports and metastatic disease. Among them, based on the presence of gain alteration, 15.2% of pts received TT. Overall survival did not significantly differ in pts with and without gain mutations (HR 1.10, 95% CI: 0.93-1.30). An exploratory analysis of 17 matched treated vs untreated pts showed better overall survival (HR 0.26, 95% CI: 0.08-0.79) for the treated ones. More than 50% of pts with gain mutation did not received TT due to drug unavailability (lack of trials, expanded access programs or off-label use) or decreasing of general conditions (38.1%). Conclusions: The selection of appropriate samples and a bidirectional pathologist-clinician collaboration is crucial to enhance the feasibility of CGP. In our wide population, a relevant proportion of metastatic gastrointestinal pts presented a druggable gene alteration with a potential improvement of outcomes in case of availability of TT. Moreover, the time of testing is important to do not lose the right window of opportunity, especially in the metastatic setting.