Species distribution models (SDMs) often use elevation as a surrogate for temperature or utilise elevation sensitive interpolations from weather stations. These methods may be unsuitable at the landscape scale, especially where there are sparse weather stations, dramatic variations in exposure or low elevational ranges. The goal of this study was to determine whether radiation, moisture or a novel estimate of exposure could improve temperature estimates and SDMs for vegetation on the Illawarra Escarpment, near Sydney, Australia. Forty temperature sensors were placed on the soil surface of an approximately 12,000 ha study site between November 2004 and August 2006. Linear regression was used to determine the relationship with environmental factors. Elevation was correlated more with moderate temperatures (winter maximums, summer minimums, spring and autumn averages) than extreme temperatures (summer maximums, winter minimums). The correlation (r 2) between temperature and environmental factors was improved by up to 0.38 by incorporating exposure, moisture and radiation in the regressions. Summer maximums and winter minimums were predominately determined by exposure to the NW and coastal influences respectively, while exposure to the NE and SW was important during other seasons. These directions correspond with the winds that are most influential in the study area. The improved temperature estimates were used in Generalised Additive Models for 37 plant species. The deviance explained by most models was increased relative to elevation, especially for moist rainforest species. It was concluded that improving the accuracy of seasonal temperature estimates could improve our ability to explain the patchy distribution of many species.