The prognostic role of spontaneous portosystemic shunts (SPSS) has been poorly investigated. To evaluate the impact of the presence of SPSS, as well as their characteristics, on the risk of decompensation. This is a retrospective cohort study of 235 advanced chronic liver disease (ACLD) patients with available imaging examination, transient elastography, and upper endoscopy. ACLD was defined as liver stiffness measurement (LSM) >10 kPa. Competitive risk analyses were performed to identify the factors associated with the main outcome. SPSS were reported in 141 (60%) of the patients. Non-viral etiology was independently associated with SPSS presence Odds-Ratio (OR): 2.743;95%-Interval-of-Confidence (IC):1.129–6.664. During a follow-up of 37 (20–63) months, SPSS were found predictors of any decompensation type Subhazard Ratio (SHR):2.264; 95%-IC:1.259–4.071, independently from a history of decompensation or high-risk-varices presence. The risk of complications was higher in patients with large (SHR: 3.775; 95%-IC: 2.016–7.070) and multiple (SHR:3.832; 95%-IC: 2.004–7.330) shunts, and in those with gastrorenal shunts (SHR:2.636; 95%-IC:1.521–4.569). The presence, size, and number of SPSS predict not only the risk of hepatic encephalopathy but that of any type of decompensation across all stages of cirrhosis. Future studies should explore the possibility of treating shunts to prevent decompensation.