Carbon fractions of tree tissues are a key component of forest carbon mass estimation. Several methods have been used to measure carbon fractions, yet no comprehensive comparison between methods has been performed. We found significant differences between carbon fractions derived from four sample preparation methods: oven-drying, vacuum desiccation, freeze-drying, and a new method that consisted of (i) not drying samples, (ii) cutting samples instead of grinding them, (iii) measuring carbon content of samples, (iv) oven-drying remaining sample material and (v) using mass measurements of remaining sample material before and after oven-drying to adjust measured carbon fraction values to an oven-dry basis (minimize the loss of carbon (MLC) method). Oven-drying, freeze-drying and vacuum desiccation resulted in lower average carbon fraction estimates than the MLC method, suggesting that they do not capture as much of the carbon present in tree tissues. Further analysis showed significant, though small, differences in carbon fractions between powdered samples and samples excised from tree core segments with a razor blade. Powdered samples were found to have lower carbon fractions than the excised samples, indicating that some carbon is lost when samples are powdered instead of cut. Utilization of the MLC method captured an average of 1.4% more carbon on an oven-drying basis than freeze-drying, the next best method. Additionally, when applied to different tree tissue types, these methods measured different volatile carbon fractions, indicating that studies attempting to quantify volatile carbon and total carbon fraction in trees should measure all tissue types present.