A general discussion is given of the problems of assigning changes in nuclear spin relaxation rates, brought about by changes in system composition, temperature, etc., to changes in correlation times and/or nucleus-lattice coupling energies. In particular the usually assumed proportionality of correlation time and viscosity is examined in detail for aqueous electrolyte solutions. It is concluded that for ions which have a marked effect on the ‘structural equilibrium’ of the water, the correlation time—viscosity proportionality is unlikely to exist. It is suggested that in favourable cases individual particle self-diffusion coefficients are likely to be related in a simple way to correlation times. This is supported by available results for aqueous potassium iodide solution.