Diffusion-ordered spectroscopy (DOSY) is a powerful method for the NMR analysis of many types of mixture without the need for physical separation, and requires only relatively standard spectrometer hardware. The principal requirements for high resolution analysis using DOSY, that the basic NMR spectrum be well-resolved and that it have good signal-to-noise ratio, pose a dilemma where multiple chemically similar species with NMR-active heteronuclei are involved. Generally the 1 H spectrum of such a mixture has good sensitivity but relatively poor chemical shift resolution, while for heteronuclei the situation is reversed. The dilemma is illustrated for the case of a mixture of cyclic dimethylsiloxanes, and the results of a range of 1 H 2D DOSY, 29 Si 2D DOSY and 1 H { 29 Si} 3D DOSY-HMQC experiments are compared. In selecting the most appropriate technique to use for a given sample it is necessary to balance conflicting requirements for speed, resolution and accuracy, and to consider the balance between systematic and random errors. Results are presented for three different concentrations of a mixture of cyclic siloxanes containing between 4 and 20 monomer units. The diffusion coefficients measured show an inverse half power law relationship with molecular mass.