This paper presents a statistical model for prediction of feedstock’s crackability (potential to generate valuable products on catalytic cracking process), based on experimental reactivity data by microactivity test (MAT - Microscale Fixed Bed Reactor) and detailed physicochemical characterization. A minimum amount of experimental tests corresponding to feed properties (typically available at refinery) is used to build a more complete description of feedstocks including chemical composition and hydrocarbon distribution. Both measured and calculated physicochemical properties are used to predict the yields of main products at several MAT reaction severities. Different well known functions correlating yields and conversion (previously tested with experimental data MAT) allows the evaluation of maximum point of gasoline yield. This point is used like a crackability index and qualitative point comparison of feedstock’s potential. Extensive feedstocks data base from Instituto Colombiano del Petróleo (ICP) with a wide range of composition were used to build the model, including the following feeds: 1. Light feedstocks - Gasoils of refinery and laboratory cuts from different types of Colombian crude oils and 2. Heavy feedstoks - Residues or feedstocks combined (blending of gasoil [GO], atmospheric tower bottom [ATB], demetallized oil [DMO] and demetallized oil hydrotreated [DMOH] in several proportions) from the four fluid catalytic cracking units (FCCU) at Ecopetrol S.A. refinery in Barrancabermeja - Colombia. The results of model show the prediction of valuable products such as gasoline for different refinery feedstocks within acceptable accuracy, thus obtaining a reliable ranking of crackability.