Long-lived trees preserve a record of environmental conditions during their lifetime in the pattern of yearly xylem widths and in changing wood density within and among the increments. Crossdated earlywood, latewood, and total ring widths, and minimum earlywood and maximum latewood densities, from three sites in the mountains of Maine, are analyzed visually and statistically to evaluate their relationships to one another and to external, environmental factors which affect the ring width and density through internal, physiological processes. Maximum density values show highest levels of similarity within and among the three site chronologies, thus implying a good degree of sensitivity to climate; minimum density values, however, showed lowest sensitivity to climate. Two biologically reasonable hypotheses concerning climate--tree growth interactions are proposed: (1) that maximum density is related to spring temperatures prior to its formation; and (2) that maximum density is related to summer water relations as the latewood forms. With the help of response function analysis, simple correlation, and multiple linear regression, these two hypotheses are tested: (1) maxmum density as a single predictor explains up to 37% of spring temperature variance; with earlywood widths at one site, 47% of spring temperature variance is explained; (2) maximum density as a single predictor explains up to 45% of summer temperature variance, 11% of summer precipitation variance, and 23% of the variance of Thornthwaite water deficit values; with total ring widths at one site, 22% of the variance of summer stream runoff is explained. Regression equations were applied to the 201- to 310-year tree-ring records to form reconstructions of these past climatic events. Independent verification testing of the reconstructions strongly validates the relationship between maximum density and spring temperature; the relationship to summer water relations is not as strongly verified, but results encourage further testin of this relationship. Results from this study may be applied both to (1) an increased understanding of relationship of climate to the formation of wood density; and (2) further development of dendroclimatology in mesic regions such as northeastern North America.