The paper addresses the problem of selecting appropriate climatic variables from readily available weather data for use in studies of species ranges and ecosystem site classification. Principal component variables and more traditional climatic variables such as heat sums are presented. To facilitate comparison of spatially distributed variables, a procedure is described for estimation of climatic statistics for areas between weather stations. The procedure is objective and provides a measure of its error. Like any regression procedure, it improves with availability of more data. Extension of climatic statistics to a 5 × 5 km grid covering the state of Michigan was used to create contour maps for a number of climatic statistics with potential relevance for plant growth. In addition, a large number of climatic statistics were summarized using principal component analysis. Separate analyses were made for winter temperature and precipitation, growing season temperature, growing season precipitation, and a combination of variables possibly related to stressful conditions. There was a high degree of correlation among many of the statistics. The correlations were due to global climatic controls and to moderation due to the Great Lakes. Principal component variables successfully presented major climatic trends. However, for ecological use they appeared to offer few advantages over more traditional climatic statistics.