Expected increase of world wide food demand requires improvement of efficient water use in agriculture in arid and semi-arid regions. The Food and Agriculture Organization (FAO) of the United Nations, the Directorate-General for International Cooperation (DGIS) of the Netherlands and many more key actors at the level of policy, research and practice are involved to obtain this improvement. However, this thesis demonstrates that there is little agreement between key actors regarding most relevant indicators for efficient water use and most effective strategies to obtain an improvement of efficient water use at the agricultural field. At field scale, present indicators and strategies are analyzed. Two typical actual fields are simulated for a single growing season using the Soil-Water-Atmosphere-Plant model (SWAP) and WOrld FOod STudies simulation model (WOFOST), calibrated against output from the Surface Energy Balance Algorithm for Land model (SEBAL). Remote sensing and model data is used to significantly reduce field work generally required in hydrological research. The obtained baseline scenarios are a plausible representation of the actual fields.Furthermore, SWAP/WOFOST allows for the simulation of various strategy scenarios. Ten different strategies for improvement of efficient water use are observed. The model output of baseline and strategy scenarios is used for computation of 13 different indicators for efficient water use. Hence, quantification of improvement of efficient water use by strategies according to possible indicators is obtained. Strategies and indicators correspond to present perceptions of key actors. The used methodology for field scale analysis is proven effective in this research. It is expected to be applicable for other regions and crop varieties. Recommendations are made concerning the methodology and future research on improvement of efficient water use in agriculture.The fields observed in this research are a general surface irrigated winter wheat field in Tadla basin Morocco and a sub surface irrigated smallholder maize field in the Lower Limpopo basin Mozambique. This thesis demonstrates that the effect of strategies is field specific. In general, a significantly larger potential for improvement is observed for the smallholder maize field. Also, trusted strategies are shown to be counter effective. Furthermore, change in efficient water use is greatly uneven and sometimes opposing by different indicators.At the winter wheat field, the target of 25% increase of the water productivity indicator used by DGIS is not met by any of the observed strategies. At the maize field this target is met by, among other strategies,elimination of irrigation. However, this also results in a 87% decrease of seasonal yield. Key actors use multiple different water productivity indicators, that are expressed in kg m¡3 and correspond to ’crop per drop’ or more vague and conceptual definitions for water productivity used at the FAO. The change to optimal seed quality at the smallholder maize field results in a water productivity increase ranging between -26 and +148% by different water productivity indicators. The -26% is obtained according to the indicator used by DGIS. This thesis demonstrates that the results from this indicator are misleading, caused by the use of biomass production in the nominator of the water productivity definition. Yield production is more representative for the desired field performance. The ’drop’ in the water productivity denominator can refer to applied irrigation water as observed in the UN Sustainable Development Goal (SDG) indicator 6.4.1, or to other water balance fluxes including evapotranspiration or transpiration. Evaluating the applied irrigation water is relevant when data is available regarding efficient use of water for other purposes than field application.This is outside the scope of this research. Evapotranspiration and transpiration provide information on the consumption of water by the observed system and by the crop. These quantities can be accurately monitored with remote sensing technologies. Therefore this thesis suggests that in arid and semi-arid regions, the water productivity indicator defined by yield divided by crop transpiration is the most relevant indicator for efficient water use to the purpose of food security. Although there is currently little agreement among key actors, the largest consensus on a relevant indicator was found for this definition. It is also demonstrated that indicators are often unclear to key actors involved in practice or at policy level and that key actors involved in research are most critical. This implies possible challenges in implementation of a single indicator for global use. In world wide monitoring of this indicator, the greatest challenge is expected in the computation of yield from biomass production for which land use classification is required. This thesis therefore also emphasizes the need for the development of methodologies that allow world wide mapping of agricultural land use.