Ecological–economic model simulations of the world food system have been used to study the impacts of historic and future liquid transport biofuel expansion on agricultural markets and the environment. Almost half of global cropland increase between 2000 and 2008 (about 8 Mha or 0.5% of global cropland) can be attributed to biofuel expansion alone. The central ‘New Policies Scenario’ of the World Energy Outlook 2011 projects an increase of conventional crop-based biofuel use from 60 Mtoe (2.5 EJ) in 2010 to annually 160 Mtoe (6.7 EJ) in 2035. Until 2020, the projected biofuel consumption provides no or little cumulative net greenhouse gas (GHG) savings as the time period is hardly sufficient to compensate for carbon losses due to over 10 Mha of additional land use conversion. By 2035, cumulative net GHG savings improve up to 2.8 Pg CO2 equivalent in a scenario with assumed higher agricultural productivity growth in developing countries. This scenario increases the developing region's competitive positions and avoids additional people at risk of hunger due to higher commodity prices caused by biofuel use. Available underutilized grassland and woodland may provide land resources suitable for nonfood energy crop production required for the second-generation biofuel conversion pathways, while causing only limited impacts on food security and biodiversity. We estimate between 246 and 475 Mha of global grassland and woodland to be agronomically suitable for industrial-scale lignocellulosic energy crop production, with achievable rain-fed yields of at least 10 tons of dry matter per hectare, with good accessibility and relatively low ruminant livestock density.