The optimal fertilizer N rate varies widely among fields and years due to variation in soil N supply and crop N demand. Soil-based diagnostic tests can be used to predict soil N supply as an aid to improved fertilizer N recommendations. Here we examine the challenges and options for predicting soil N supply in rain-fed potato production in eastern Canada. Soil N supply was assessed as the plant N uptake in unfertilized plots. Most residual nitrate is leached from the root zone over the winter period and consequently spring soil mineral N primarily reflects early season soil N mineralization. Spring soil nitrate is highly correlated with soil N supply in an individual growing season, but is not a reliable predictor of soil N supply. A number of indices of soil mineralizable N have been evaluated and several of these show promise. Practical application of these indices is complicated by climatic variability among growing seasons. A simple first-order kinetic model was shown to be effective in prediction soil N supply bt taking climatic variation into account. However practical application of such kinetic models is limited by the effort required to obtain the necessary input parameters. Promising areas for future research are identified.