Methods for examinations of diffusion of large molecules of the size of fatty acids or triglycerides were developed for whole body MR units. Samples of aliphatic molecules were examined to study the influence of chain length. Feasibility under in vivo conditions was tested on lard samples at 37 degrees C and on human subjects Three stimulated echo sequences with maximum b-values of 2000 s/mm(2), 20000 s/mm(2), and 80000 s/mm(2) were used to assess a wide range of mobility. Sequence timing was optimized to minimize relaxation losses of fatty tissue. Apparent diffusion coefficients (ADC) were determined from five spectra with different diffusion weighting. In-vitro experiments were performed on butanol, decanol, and oleic acid to study the influence of chain length. In vivo conditions were mimicked using lard at 37 degrees C representing a composition of substances of various chain lengths. Subcutaneous fat and tibial bone marrow were studied in three healthy volunteers. ADC of muscular lipids of the lower leg was determined in two subjects. ADC values of pure aliphatic substances were in the range between 3.2 x 10(-5) mm(2)/s for oleic acid and 37.8 x 10(-5) mm(2)/s for butanol. In vivo investigations revealed ADC values of 1.11-1.24 x 10(-5) mm(2)/s for tibial bone marrow and 1.21-2.05 x 10(-5) mm(2)/s for subcutaneous fat. Diffusion coefficients of extra- and intramyocellular lipids were 1.83-3.65 x 10(-5) mm(2)/s and 2.22-3.60 x 10(-5) mm(2)/s, respectively. The proposed technique enables determination of ADC values of relatively large molecules and of lipid tissue compartments under in vivo conditions. Diffusion properties in several human lipid compartments are reported for the first time. Incoherent voxel motion influences the in vivo results to an unknown degree because of high motion sensitivity. In vitro experiments revealed ADC values depending on the chain length of the substances, indicating a residual dependence of measured ADC's on sequence timing.