Malaria remains a public health issue, with about 228 million cases and 500.000 deaths annually.1 Evolution of parasite strains resistant to all known classes of antimalarial drugs urges the need for novel lead compounds, including to block transmission.2 Aiming to discover new antimalarial drugs, we developed robust and predictive machine learning models for predicting transmission-blocking activity of compounds against Plasmodium berghei. Initially, a high throughput screening collection of 6636 compounds with experimental transmission blocking data was curated and standardized according to best practices of Fourches and co-workers.3 Then, categorical machine learning models combining ten fingerprints (AtomPair, Avalon, Dragon, FeatMorgan, Layered, MACCS, Morgan, Pattern, RDKit, and Torsion) with two machine learning methods (Random Forest and Naïve Bayes) were developed following the best practices of predictive modeling.4 As a result, we have obtained statistically predictive models, with CCR, Sensitivity, and Specificity values ranging between 0.79–0.84. Subsequently, the most predictive model (combining Random Forest and Torsion fingerprints) was used in a virtual screening campaign along with two other models (asexual stages of P. falciparum W2 and 3D7 strains) previously developed in our laboratory.5 Using these models, we prioritized 40 compounds for further in vitro testing against asexual and sexual stages of Plasmodium spp.