Abstract Field assessments of crop water use efficiency (WUE) are resource‐consuming since they require simultaneous assessment of the total amount of water assimilated by crops for biomass and/or grain production. Alternative methods exist, such as estimating the carbon isotopic ratio ( 13 C/ 12 C) of the crop's leaf, aboveground biomass, or grain samples. There is limited information on the determinants of the accuracy of carbon isotopes in estimating water use efficiency between crop types and environments. Therefore, this study aimed to evaluate the extent to which the estimation of the 13 C/ 12 C ratio in crop parts constitutes an accurate proxy of WUE, globally. Data on observed WUE (WUE obs ) were collated involving 518 experiments conducted worldwide on major cereals and legumes and compared with WUE estimates (WUE est ) from carbon isotopes. The mean WUE obs among all experiments was 3.4 g L −1 and the mean absolute error (MAE) was 0.5 g L −1 or 14.7% of WUE obs , corresponding to accurate predictions at p < 0.05. However, the percentage mean absolute error of observed water use efficiency (%MAE) estimated from grains was 3.6 ± 11.5%, which was lower than the %MAE from aboveground biomass collected at harvest (3 ± 22.8%). In addition, the %MAE increased from 1.1 ± 5.1% for soybean, 1.6 ± 7.2% for maize, 1.2 ± 8.6% for rice, 1.8 ± 12.1% for groundnut, 2.1 ± 14.3% for cowpea, 2.3 ± 16.2% for bush bean, 1.8 ± 19.9% for wheat, 2.2 ± 21.4% for barley to 6.3 ± 39.3% for oat, with only the latter corresponding to significant errors. WUE est were, in all cases, unbiased but slightly overestimated from 0.8% (maize) to 15.4% (oat). The accuracy in estimating WUE significantly decreased with the increase in soil clay content, with sand, showing a positive correlation of 0.3 with %MAE, but negatively correlated with the silt content ( r = −0.4). Furthermore, a multivariate analysis pointed out a tendency for prediction errors and bias to increase with the decrease in WUE obs and air temperature. Using carbon isotopes for estimating crop WUE thus appeared reliable for all crops and world environments, provided grain samples are considered. The technique tended to perform better under high WUE conditions, such as those generally found in maize and soybean cropping systems. The identified factors that affect the accuracy of using carbon isotopes in measuring WUE provide valuable insights for water resource management and sustainable crop production. These findings contribute to the ongoing discourse on water conservation strategies in agriculture, offering a basis for decision‐making in crop improvement programs. Implementing the recommended practices from this study can potentially improve yield gains and promote resilient and sustainable agricultural systems in the changing environmental circumstances. Further research should investigate the mechanisms that cause low accuracy of the isotopic technique using aboveground biomass and under arid and cool environments.