This article describes how the analysis and statistical treatment of data, spontaneously collected or generated through a set of designed experiments, resulted in a better understanding of the process and the identification of which and how major variables influence its behavior and performance at SOPORCEL Pulp and Paper Plant. Monthly averages of main process variables for the past 5 years of manufacturing were employed in order to perform an analysis and build an empirical model. The model includes 6 independent variables and is able to predict the Pulp Shopper Riegler Degree, degrees SR, before refining, explaining around 70% of its variability. Full factorial statistically designed experiments in the production line were used to study the relationships between refining conditions and a number of different daily averages for Refined Pulp Characteristics. From the results obtained, it was possible to find prediction models that are also able to explain around 70% of the variability for most of the responses.