Liver transplantation (LT) selection models for hepatocellular carcinoma (HCC) have not been proposed to predict waitlist dropout due to tumor progression. This study aimed to compare the alfa-fetoprotein (AFP) model and other pre-LT models in their prediction of HCC dropout. A multicenter cohort study was conducted in 20 Latin American transplant centers, including 994 listed patients for LT with HCC from 2012 to 2018. Longitudinal tumor characteristics and patterns of progression were recorded at the time of listing, after treatments and at last follow-up over the waitlist period. Competing risk regression models were performed, and the model's discrimination was compared by estimating Harrell's adapted c-statistics. HCC dropout rate was significantly higher in patients beyond [24% (95% CI 16-28)] compared to those within Milan criteria [8% (95% IC 5-12%); P<.0001], with an SHR of 3.01 (95% CI 2.03-4.47)], adjusted for waiting list time and bridging therapies (c-index 0.63 (95% CI 0.57;0.69). HCC dropout rates were higher in patients with AFP scores >2 [adjusted SHR of 3.17 (CI 2.13-4.71)], c-index of 0.71 (95% CI 0.65-0.77; P=0.09 vs. Milan). Similar discrimination power for HCC dropout was observed between the AFP score and the Metroticket 2.0 model. In patients within Milan, an AFP score >2 points discriminated two populations with a higher risk of HCC dropout [SHR 1.68 (95% CI 1.08-2.61)]. Pre-transplant selection models similarly predicted HCC dropout. However, the AFP model can discriminate a higher risk of dropout among patients within Milan criteria.