to accurately model soil temperatures on the boreal plain, factors that influence fine-grained soils during the rapid cooling period must first be identified. the effects of air temperature, soil moisture and snow depth were quantified at 0.1 and 0.5 m depths for 14 sites encompassing five treatment types: three upland burned, three upland harvested, three upland conifer, three upland deciduous and two wetland. in the absence of snow from september to october at the 0.1 m depth, air temperature was identified as the most important parameter, explaining approximately 70% of the variation in soil temperature for upland and wetland sites. at the same depth in the presence of snow from november to december, soil moisture was more important. at a deeper soil depth (0.5 m), soil moisture was identified as the most important parameter regardless of snow cover, explaining from 63 to 91% of the variation in soil temperatures for upland and wetland sites. the presence of snow was a significant factor influencing soil temperatures, but snow depth was not. further, the soil temperature algorithms of swat were tested using one site of each treatment type at 0.1, 0.5 and 1.0 m depths. the algorithms utilized by swat were able to reproduce seasonal trends in soil temperatures adequately for the spring, summer and autumn seasons, with only a slight increase in the lag coefficient parameter. during winter months, the swat algorithms tended to predict soil temperatures that were consistently lower than measured data. further development to the swat soil temperature algorithms is required to represent better the important insulating effect of snowpack.