Knowledge about the triglyceride (fat) 1H spectrum enables quantitative determination of several triglyceride characteristics. This work describes a model-based chemical shift imaging method that separates water and fat signal and provides maps of three triglyceride quantities: fatty acid carbon chain length, unsaturation degree, and polyunsaturation degree. The method was validated by imaging a phantom containing 10 different oils using 1.5 and 3.0 T clinical scanners, with gas–liquid chromatography as reference. Repeated acquisitions demonstrated high reproducibility of the method. Statistical tests of correlation and linear regression were performed to examine the accuracy of the method. Significant correlation was found at both field strengths for all three quantities, and high correlation (r2 > 0.95) was found for measuring unsaturation degree and polyunsaturation degree. Feasibility of the method for in vivo imaging of the thigh was demonstrated at both field strengths. The estimates of unsaturation degree and polyunsaturation degree in subcutaneous adipose tissue were in agreement with literature values, while chain length appeared over estimated. The method has potential use in large-scale cross-sectional and longitudinal studies of triglyceride composition, and its relation to diet and various diseases. Magn Reson Med, 2012. © 2012 Wiley Periodicals, Inc.