We studied the spatial auto- and covariation in the growth of 264 trees and their root zone soil sodicity in a fine loamy natric Haplustalf sampled from four depths at 30-cm increments by the regionalized variable theory. Stochastic heterogeneity was modeled after the deterministic trend was removed nonparametrically. Variability in the soil pH, sodium adsorption ratio (SAR), and DTPA-extractable Na and in tree height measured 1, 2, and 3 yr after planting was anisotropically spatially structured. Interdependence between soil sodicity and tree growth was also anisotropically spatially distributed. Autovariability, and especially cross-variability, was relatively higher in the northwest to southeast direction than in the northeast to southwest direction. Cauchy-Schwarz's condition of inequality of cross-variogram functions was always satisfied. Predictability of variation of the two intrinsic random fields was validated by computing various statistics of the values kriged by jackknifing and actually measured observations. Possibilities of improving the precision of evaluating afforestation techniques through spatially sensitive statistical procedures for realizing better management of sodic soils are discussed.