Researchers reviewed 24 studies with 5,197 patients, yielding 32 models for predicting the risk of hepatic encephalopathy (HE) after TIPS. The incidence of HE ranged from 19.9% to 46.6%. The models had good discriminatory power with AUCs in the range of 0.64–1.00, with 30 models having AUCs above 0.70 and 22 above 0.80. The meta-analysis showed a pooled AUC of 0.815 (95% CI: 0.780–0.849). Consistent risk predictors (p < 0.01) were older age, diabetes, higher Child-Pugh score or class, increased ammonia, and higher portal to splenic vein diameter ratio. The PROBAST evaluation revealed a high level of bias, especially in the model analysis. The authors recommend multicenter studies with harmonized definitions, rigorous analysis and external validation for the future.