Leaf size and shape are selected by climate and are strongly correlated with climatic variables. Accordingly, fossil dicotyledonous leaves are considered to be among the most reliable indicators of terrestrial paleoclimates. However, the methods used have much potential for improvement. All currently rely on discrete, usually binary characters, but leaf morphology is better evaluated and more reproducibly measured using continuous variables. Digital leaf measurement offers the possibility of continuous, reproducible variables, with the potential to improve paleoclimate estimates as well as the ecological and evolutionary understanding of leaf form. Results are reported for five variables measured on 238 leaves representing the woody species from three living test floras—two temperate and one humid tropical. The variables are: shape factor (an area:perimeter ratio standardized to a circle); the ratio of feret diameter (diameter if leaf’s area is contained in a circle) to major axis length; the ratio of tooth area to leaf area; tooth count; and tooth count standardized to leaf perimeter. It is well known that increased mean annual temperature is correlated with a high percentage of untoothed species. By extension, increasing temperature is expected to be associated with relatively undissected, more circular leaves (i.e., with high shape factor, low tooth-area:leaf-area ratio, and low tooth count). These trends were observed in the test floras, both between the tropical and the temperate sites and between the two temperate sites. Shape factor was the variable that resolved the three sites most significantly. The five variables show promise for testing on an expanded set of floras sufficient to develop statistical models for paleoclimatic studies.