Abstract The information provided in publications on water-related agronomic trials and irrigation interventions is often too limited to compare values of water productivity (WP), i.e. the ratio between produced plant biomass and the amount of water used for that production, from different years, regions, etc. in a meaningful way. In this article, we show with the help of simulation models how WP-values are affected by different definitions of the numerator and denominator, environmental circumstances, such as climate, year and sowing date, and crop characteristics. In many cases, this resulted in 10–25% change in the WP-values, and sometimes even more. A minimum dataset is formulated that will make normalization and comparison of different WP-values easier. Most of these data are known by those who execute experiments, and we recommend strongly that these are reported in the future. Simulation models are excellent tools to explore the limitations and opportunities for increasing WP, provided they are well calibrated and validated for biomass, soil water availability, and ET. Such a balanced estimation of the “crop” and the “drop” requires an improved cooperation between hydrologists and agronomists. Comparison of actual WP (E)T and simulated maximum WP (E)T for the same environmental conditions does show the scope for increasing WP (E)T and other WP-values. Since WP-values are ratios, the production level on a hectare basis should be given besides WP. When we try to find an optimum combination of production per hectare and production per m 3 irrigation water, we will be able to produce “more food with less water”.