Abstract: The results of experiment“3414”of rapeseed in soil-testing formulated fertilization during the period of 2006 to 2008 in Zhejiang Province were analyzed. The results indicated that the yield of rapeseed ranged from 1 665 kg·hm^-2 to 3 200 kg·hm^-2 in the optimum treatment (N₂P₂K₂), and the yield was lower in soils with lower pH and available K. The average yield\|increasing rates under different fertilization conditions were 117.5% for N fertilizer, 45.1% for P fertilizer and 7.3% for K fertilizer, and the average increasing yield per unit of N, P₂O₅, K₂O were 4.7, 7.0, 1.7 kg, respectively. There was great difference in ferti lizer efficiency among different soil types. The correlation analysis of relative yields in the treatments of nutrient-deficiency and soil testing values showed that the relative yields in the treatments of N-deficiency showed significant correlation with soil organic matter and total N contents, which could be used as the indicator of N fertilization recommendation. On the contrary, poor correlations were found between the relative yields in the treatments of P or K deficiency and soil available P or K, i.e. the significant correlations were only found in 50% investigated sites for P and 70% investigated sites for K, which implied that there were some limitations in soil P or K analysis methods when used for fertilization recommendation. Thus different indicator systems should be made according to different soil types.

Key words: rapeseed, NPK fertilizer effect, soil testing value, correlation