Predictive statistical models are widely used in research development to represent linear and non–linear trends that occur in natural phenomena, generally associating the effect caused by multiple adjustable parameters on measurable experimental results. One of the greatest benefits of these models is their application in predictive analysis, since this allows to facilitate decision–making at industrial level in short time, and therefore they are widely used in sectors such as petrochemical industry. In this field, predictive statistical models are generated from experimental results for dependent variables such as materials corrosion rate. In these cases, the models are established based on dependent variables such as: chemical composition of corrosive medium; exposure time; system temperature and type of exposed material, among others. For this reason, in present investigation a statistical model was determined for corrosion rate of AISI 316 steel exposed to a Colombian heavy crude oil as a function of system temperature and exposure time. Development of this statistical model allows its industrial implementation as a tool for prediction of AISI 316 steel sulfidic corrosion rate in a transfer line used for heavy crude oil refining in temperature and exposure time ranges commonly present in distillation units.