Abstract: In order to rapidly obtain growth information of the rapeseed in field, the models of leaf area and fresh weight of a single plant should be built on the growth conditions under different fertilization levels and sowing ways. Plot experiments with different fertilization levels and sowing ways were conducted in two consecutive rapeseed planting seasons in 2013-2015. The length, width, area and fresh weight of each leaf of the sampled plants were measured at six-leaf stage and bolting stage, in which the leaves of rapeseed were of the most morphological variation. The analysis of variance was used to analyze the influence of different planting patterns and fertilization levels on rapeseed leaf. And the Levenberg-Marquardt and Universal Global Optimization methods were used to optimize the length-width function model to estimate leaf area (LA) and fresh weight (FW). The correlations among leaf area, fresh weight and leaf width was more significant than those among leaf area, fresh weight and leaf length under different fertilization levels and sowing ways at six-leaf stage and bolting stage. Variation of LA and FW at bolting stage was greater than those at six-leaf stage. Moreover, there were significant variance of LA and FW under different planting patterns and N treatments levels. No significance was found in cross verification. LA and FW were estimated well with leaf width (R²=0.89 and RMSEP=32.40 cm² for LA model and R²=0.84 and RMSEP=2.54 g for FW model). The validation of models (R²=0.97 and RMSEP=12.92 cm² for LA model, and R²=0.94 and RMSEP=0.86 g for FW model) showed that length-width power function models were of the highest precision compared with other models. Under different growth conditions, the leaf width linear model was promising to rapidly estimate leaf area and fresh weight of a single rapeseed plant, while length-width power function models could be used to accurately estimate leaf area and fresh weight in large field with minor influence of leaf morphology and growth stages.