to estimate the level and pattern of variation in frost hardiness, artificial freezing tests of 64 provenances of jack pine were conducted. the provenances originated from northern ontario. seedlings of the provenances were grown in a uniform environment in a shade house. current-growth needles were collected in fall during three consecutive years , 1988 to 1990, and in mid-summer in 1990. three test temperatures and a control were used for all freezing trials. temperatures ranged from - 19° c to -1° c and duration varied from three to one hours. freezing injury was evaluated visually. two way anova indicated statistically significant provenance and provenance x temperature interactions. these results suggested that the tested jack pine provenances exhibited genetic variation in their development of frost hardiness and implied a certain risk in transferring seed from one environment to another. differentiation among provenances could not be detected during early august 1990. regression analyses examined the associations between various degrees of injury and climatic gradients. these analyses suggested that several selective forces, including precipitation and temperature, were partially responsible for differentiation among the tested provenances. principal component analysis (pca) of the data generated three significant principal components which accounted for approximately 60% of the total variation. regression of pca scores against climatic gradients also reflected adaptive variation. however, a number of provenances which originated from regions with low temperatures and very short frost-free periods showed no higher levels of frost hardiness than provenances from areas with longer frost-free periods and higher temperatures. low levels of consistency were found among the different trials. possible reasons for the observed inconsistencies were assumed to be i) weaknesses of the scoring technique, ii) the random effect of supercooling and, iii) the uneven distribution of temperature in the freezer.