It is well known that coke has a major importance on the final cost of the Hot Metal and besides this fact its properties are closely connected with the performance of blast furnaces. Therefore, seeking a type of coke compatible with the blast furnace technical requirements and at the lowest possible price may become a hard challenge if neither a prediction tool is available to accomplish a correct selection of coals nor previous test were made on coke test facility. Among the properties used by the Blast Furnace specialists to qualify the coke, it can be highlighted CRI - Coke Reaction Index and CSR - Coke Strength after Reaction with CO2, because they represent a direct measure of the Boudouard reaction – solution loss, and its association with the coke strength, respectively. Based on them it is possible to predict the consumption of this raw material in the Blast Furnaces and, due to its high influence on the operational costs, measure its impact on the overall process economy. The Coke Plant unit at thyssenkrupp Companhia Siderurgica do Atlantico - TKCSA follows the coking process known as heat recovery and its concepts differ from the traditional one, by product. This unit uses its own coke test facility, specifically designed by the company s experts to simulate the operational conditions at the coke plant industrial furnaces. However, the coke test facility still required further specific experimental study capable of making a reliable comparison analysis between the cokes produced in it with the one from the industrial coke batteries. The main goal of this dissertation is to develop model to predict the CRI - Coke Reaction Index and CSR - Coke Strength after Reaction with CO2, through the coking process of coal blends, generated by different methodologies, charged on the coke test facility statistically compared with the ones from the industrial Coke plant at thyssenkrupp Companhia Siderúrgica do Atlântico - TKCSA. The prediction of CSR represented when generated through the coke test facility showed good adherence with the values obtained with the industrial scale. Therefore, the current procedures used on the developed model has proved to be adequate and may be used as a predicting tool to help the purchase of coals.