Abstract. Croplands cover about 12% of the ice-free terrestrial land surface. Compared with natural ecosystems, croplands have distinct characteristics due to anthropogenic influences. Their global gross primary production (GPP) is not well constrained and estimates vary between 8.2 and 14.2 Pg C yr⁻¹. We quantified global cropland GPP using a light use efficiency (LUE) model, employing satellite observations and survey data of crop types and distribution. A novel step in our analysis was to assign a maximum light use efficiency estimate (ϵ^*_GPP) to each of the 26 different crop types, instead of taking a uniform value as done in the past. These ϵ^*_GPP values were calculated based on flux tower CO₂ exchange measurements and a literature survey of field studies, and ranged from 1.20 to 2.96 g C MJ⁻¹. Global cropland GPP was estimated to be 11.05 Pg C yr⁻¹ in the year 2000. Maize contributed most to this (1.55 Pg C yr⁻¹), and the continent of Asia contributed most with 38.9% of global cropland GPP. In the continental United States, annual cropland GPP (1.28 Pg C yr⁻¹) was close to values reported previously (1.24 Pg C yr⁻¹) constrained by harvest records, but our estimates of ϵ^*_GPP values were considerably higher. Our results are sensitive to satellite information and survey data on crop type and extent, but provide a consistent and data-driven approach to generate a look-up table of ϵ^*_GPP for the 26 crop types for potential use in other vegetation models.