This study aims to establish a procedure for handling co-products in life cycle assessment (LCA) of a typical sugar cane system. The procedure is essential for environmental assessment of ethanol from molasses, a co-product of sugar which has long been used mainly for feed. We compare system expansion and two allocation procedures for estimating greenhouse gas (GHG) emissions of molasses ethanol. As seen from our results, system expansion yields the highest estimate among the three. However, no matter which procedure is used, a significant reduction of emissions from the fuel stage in the abatement scenario, which assumes implementation of substituting bioenergy for fossil-based energy to reduce GHG emissions, combined with a negligible level of emissions from the use stage, keeps the estimate of ethanol life cycle GHG emissions below that of gasoline. Pointing out that indirect land use change (ILUC) is a consequence of diverting molasses from feed to fuel, system expansion is the most adequate method when the purpose of the LCA is to support decision makers in weighing the options and consequences. As shown in the sensitivity analysis, an addition of carbon emissions from ILUC worsens the GHG balance of ethanol, with deforestation being a worst-case scenario where the fuel is no longer a net carbon saver but carbon emitter. (C) 2011 Elsevier Ltd. All rights reserved.