site index of jack pine {pinus banksiana lamb) measured on 76 plots in northeastern ontario was related to features of soil and topography using multiple regression. site index at breast height age 50 years (sibhso) was used as the dependent variable, and 119 soil and topographic values were used as independent variables. regression equations were imprecise using all 76 plots. when separate equations were computed for bedrock-moraine, glaciofluvial, and moraine landforms, precision was much greater with r2 values of 0.78, 0.51 and 0.60, respectively. the final bedrock-moraine equation consisted of slope percent, thickness of the b horizon, and percent stones in the top 25 cm of the soil profile. the glaciofluvial equation consisted of depth to average rooting, depth to moisture restricting layer, and percent silt in the b horizon. the moraine equation consisted of depth to maximum rooting, pore pattern, percent sand and percent silt in the bc horizon. the northeastern ontario plots were combined with schmidt and carmean's (1988) 131 plots. new regressions based on the pooled data had r2 values of 0.84, 0.55, 0.37, 0.57, and 0.24 for bedrock-glaciofluvial, bedrock-moraine, glaciofluvial, lacustrine, and moraine landforms, respectively. these analyses produced valid jack pine soil-site equations for the combined bedrock and lacustrine landforms in northeastern and north central ontario; but equations combining data for glaciofluvial and moraine landforms were imprecise. the northeastern and north central ontario plots were pooled with 16 plots from northwestern ontario. new regressions based on the pooled data had r2 values of 0.22, 0.47, and 0.17 for glaciofluvial, lacustrine, and moraine landforms, respectively. these analyses resulted in equations that had unacceptably low precision. the failure to compute acceptable soil-site equations was attributed to different soil and topographic variables influencing the height growth of jack pine in northwestern ontario.