Evaluating the effects of management practices on soil physical and chemical properties would be valuable to explain field-level variability in crop production. A 23-year-old experiment on a Muscatune soil (fine-silty, mixed, superactive, mesic, Aquic Argiudolls) in Illinois with five N rates [0 (N 0), 70 (N 1), 140 (N 2), 210 (N 3) and 280 (N 4) kg N ha −1] and two cropping systems [continuous corn ( Zea mays L.) (CC), and corn–soybean ( Glycine max (L.) Merr.) rotation (CS)] was evaluated. Specific objectives were to: (i) evaluate the effects of long-term N fertilization and cropping systems on field level changes in soil physical and chemical properties and crop yield, (ii) identify the most responsive soil physical and chemical properties to N fertilizer and crop management, and (iii) investigate the relationship between the selected soil properties and crop yield. Soil was collected in May 2004 to 30 cm depth and 20 soil physical and chemical properties were measured. The univariate analysis indicated that 14 soil properties were significantly influenced by at least one treatment effect (crops, N or crops × N). Due to multicollinearity among soil properties, principal component analysis (PCA) was used to group correlated properties, resulting in five soil properties such as soil organic carbon stock (OC stock), mean weight diameter (MWD), soil C:N ratio, exchangeable potassium (K +) and gravimetric moisture content ( ω). Finally, the multiple regression analysis performed between PCA derived soil properties and corn and soybean yields retained all the representative soil properties from PCA except ω as yield predictors for corn ( P < 0.001, R 2 = 0.39) from CC system, whereas none of the soil properties were significantly related to corn and soybean yields from CS system. The soil properties most influenced by long-term N fertilization of continuous corn were successfully identified with PCA and multiple regression. The insignificant relationship between corn and soybean yields from CS system and PCA derived soil properties might be due to the lack of response of soybean to N fertilization. This study shows the integrated use of multivariate and regression analyses in identifying yield determining soil properties by eliminating the multicollinearity among soil properties.