This study investigates a new framework that a linear fractional programming problem is subject to fuzzy relational equations with max-product composition. Three folds are presented. First, some theoretical results are developed to optimize such a linear fractional programming problem based on the properties of max-product composition. Second, the results are adopted to reduce the feasible domain. The problem can thus be simplified and converted into a traditional linear fractional programming problem. Third, a procedure is presented to solve this optimization problem without looking for all potential minimal solutions. Numerical examples are provided to illustrate the procedure.