Brown sugar mills are characterized by their high heterogeneity regarding producers´ sociodemographic features, mills productive structure, resource endowment, technology use and economic performance. In order to test and validate a methodology to determine homogeneous groups of brown sugar mills, a random sample survey of 43 mills was researched at Ocamonte, Santander. To accomplish research´s main objectives a sequential multivariate statistic approach was applied (namely, multiple correspondence analysis with optimum scaling, cluster analysis and discriminant analysis). There were found four well differentiated types of mills representing three brown sugar production systems namely: a marginal (Type I) system encompassing 65% of mills in the sample. Another traditional improved (Type III) system involving 7% of mills. Finally, an advanced (Type II and IV) system embracing 28% of mills. Improving the productive efficiency of these mill types requires different sorts of policy interventions.