height growth patterns and site index were studied using stem analyses taken from dominant and codominant trees growing on 109 plots located in mature, natural, fully stocked evenaged stands of jack pine {pinus banksiana lamb.) in the thunder bay area. the observed height/age data were modeled using several nonlinear biological growth models: richards growth model (1959); a modified weibull function; and an expansion of the richards model proposed by ek (1971). height growth patterns of jack pine varied with level of site index, being more curvilinear as level of site index increased. height growth patterns were similar for jack pine growing on glacialfluvial sands, on moraines, on lacustrine soils and shallow to bedrock soils. analyses showed that site index curves were more precise when based on breast height age instead of total age of the trees. height growth curves, site index curves and a site index prediction equation were calculated from the jack pine stem analyses data. a modification of the chi-squared distribution was used for testing the accuracy of the site index curves and prediction equation. the accuracy of the computed curves was tested using independent stem analyses data from 32 additional confirmation plots. comparisons with this independent data showed very close agreement; the 95% prediction intervals calculated for the site index curves and site index prediction equation using independent data are -0.17 ± 0.89 m and -0.20 ± 1.14 m respectively. comparison between plonski’s (1974) formulated site index curves for jack pine and the site index curves produced in this study indicate differences in predicted heights at ages greater than index age (50 years), but no differences younger than index age. plonski’s site index curves showed lower predicted heights for each level of site index at ages greater than 60 years.