The advent of applicators having the capacity to adjust rates of fertilization as they move across fields has created a need to subdivide fields into nitrogen (N) management units, or areas of soil that should receive a common rate of N. This dissertation explores methods for using yield monitors on combines, The Global Positioning System (GPS), remote sensing, and geographic information systems (GIS) to subdivide fields into N management units. Fertilizer treatments were applied in replicated strips 4.5--6 m wide and >500 m long across fields having several soil types. Combines with yield monitors harvested each strip as a single swath. Geographic information systems were used to divide the field into 28 m2 grid cells and yield responses to treatments were calculated from the appropriate paired cells. Cells within this grid formed a population of yield responses for the whole field or for any subdivision. The field was subdivided into possible management zones by using soil survey maps or spatial pattern in light reflectance from crop canopies. Only when subdivisions resulted in yield responses great enough to pay for the treatment were the subdivisions considered different management units. Analysis showed that yield responses great enough to pay for the treatment were usually statistically significant. The experimental precision attained was considerably better than obtained in conventional small plot fertility trials. Remote sensing of canopy reflectance could identify small areas that differed substantially from the surrounding soil. This information can be used to correct and add important details to soil survey maps and maps of N management units. An important advantage of this method is that populations of yield responses are characterized for areas of soil having defined ranges of heterogeneity and, therefore, results of experiments can be used to make scientifically defensible N recommendations for areas of soil having the same defined range of heterogeneity. The major advantage, however, is that farmers can conduct trials in their fields at negligible cost and they can see the results.