Root observations in situ with a rhizotron camera enabled us to compare the performance of apple (Malus ×domestica Borkh.) trees on 3 rootstock clones planted in a New York orchard with a history of apple replant disease. Visual observations were conducted in situ at monthly intervals during 2 growing seasons through minirhizotron tubes for trees grafted onto 3 rootstocks: M.7 (M.7), Geneva 30 (G.30), and Cornell-Geneva 6210 (CG.6210). There were 3 preplant soil treatments (fumigation, compost amendment, and untreated checks) and 2 tree planting positions (within the old tree rows or in the old grass lanes of the previous orchard at this site). Preplant soil treatments and old-row versus grass-lane tree planting positions had no apparent influence on root systems, whereas rootstock clones substantially influenced root growth and demography. New root emergence was suppressed during the first fruit-bearing year (2004) on all 3 rootstock clones compared with the previous nonbearing year (2003). A root-growth peak in early July accounted for more than 50% of all new roots in 2003, but there was no midsummer root-growth peak in 2004. The median lifespan for roots of CG.6210 was twice that of G.30 and M.7 in 2004. Also, CG.6210 had more roots below 30 cm depth, whereas M.7 had more roots from 11 to 20 cm depth. Trees on CG.6210 were bigger, yielded more fruit, and had the highest yield efficiency in the third year after planting compared with trees on G.30 and M.7 rootstocks. Crop load appeared to inhibit new root development and changed root-growth dynamics during the first bearing year, with a resurgence in new root growth after fruit was harvested in October 2004. Rootstock genotype was the dominant influence on root lifespan and distribution in this study, whereas preplant soil fumigation, compost amendments, and replanting positions had little apparent impact on root characteristics despite their influence on above-ground tree growth and yield.