Wheat and maize are two major food crops in China. Conventional fertilizer recommendations result in higher than necessary costs to farmers and increased environmental pollution. It is essential to quantitatively estimate optimal fertilizer requirements to alleviate the problems of the two crops in China. The QUEFTS (QUantitative Evaluation of the Fertility of Tropical Soils) model was used to estimate region-specific nitrogen (N), phosphorus (P) and potassium (K) requirements as well as fertilizer applications needed to realize target yields of wheat and maize. Data of field experiments with different fertilization treatments of various regions in China during the years of 1985-1995 were used to calibrate the QUEFTS model for both wheat and maize. Minimum and maximum internal nutrient efficiencies (kg grain kg−1) for the model were estimated at N (25 and 56), P (171 and 367), K (24 and 67) for wheat, and N (21 and 64), P (126 and 384), K (20 and 90) for maize. The model suggested a linear increase of grain yields for scenarios with nutrient contents of 24.6, 3.7 and 23.0 kg N, P and K per 1000 kg of wheat grain and 25.8, 4.3 and 23.1 kg N, P and K per 1000 kg of maize grain. These results suggest that the average N: P: K ratio in the plant dry matter is about 6.7: 1: 6.2 for wheat and 6.0: 1: 5.4 for maize. Relationships between internal N, P and K levels and soil properties were established and relationships between the recovery efficiencies of applied fertilizer - N, P and K were found. Running the calibrated QUEFTS model with observed field data produced a good fit between predicted and observed data. It was concluded that the calibrated QUEFTS model could be a useful tool for improving fertilizer recommendations for wheat and maize in China.