The relationship between net phosphorus (P) additions or removals and soil test P (STP) changes has rarely been investigated on a landscape basis. Our objective was to characterize the relationship between soil properties and the response of STP values to P fertilizer additions and crop removals over a six-year P and nitrogen (N) fertility experiment in a corn-soybean rotation. Two 16 ha N and P fertility trials were established in the fall of 2001 on south-central Minnesota Mollisols. Annual soil testing of 15 m x 9 m grid cells within each field revealed significant responses of mean STP values to P fertilizer treatments. Corn and soybean yield was also measured annually in 15 m x 9 m grid cells. Mass-balance calculations of fertilizer P additions and crop P removals indicated substantial spatial variability in the response of STP to net P addition or removal. Rates of decline in STP under net P removal were directly related to initial STP values. Rates of increase in STP under net P addition were significantly related to pH at both sites, and to N mineralization index at one site. These results have implications for improving variable-rate P recommendations to achieve critical STP values.