Abstract Mechanized sugarcane green harvesting has been increasingly adopted in Brazil, rising straw availability at the field. Computer-aided tools are used to predict the optimum conditions for straw recovery systems. This study aims at developing a model-based method for optimization of sugarcane straw recovery costs and internal rate of return (IRR), as a function of sugarcane productivity, straw recovery fractions and transport distances considering two recovery systems: integral harvesting (IS) and baling system (BS). A simulation procedure using the Virtual Sugarcane Biorefinery (VSB), according to a Central Composite Design (CCD) is used in this study. The scenarios were based on an autonomous ethanol plant, with milling capacity of four million tons per year. The influence of these agricultural parameters on the sugarcane straw recovery costs and internal rate of return was evaluated through this approach, where the CCD was used for the development of an empirical model for optimization as well as a statistical evaluation of results. An optimized IRR (26.3%) was obtained for integral system, with low transport distance (20 km), maximum sugarcane stalk productivity (100 t ha−1) and maximum recovery fraction (70%). The same conditions lead to higher IRR (26.2%) for baling system. Results showed that the IS promotes the highest reduction of the agricultural components of straw recovery cost and the optimum IRR considering the vertical integration of sugarcane biorefineries was observed by adopting this system.