The calculation of dynamic laser-light scattering by dilute suspensions of Brownian particles is reviewed. It is shown that present theories of diffusion can provide approximations for the autocorrelation of the intensity of the scattered light that are only uniformly accurate for correlation times up to order (Do k2)-' where Do is the diffusivity of a single particle and k is the scattering wave vector. The meanings of, and connections between, down-gradient, self-and tracer diffusion for both short and long times are established and it is shown how these may be inferred from light-scattering experiments for optically monodisperse and polydisperse systems. For dilute systems, equations giving the time evolution of the intermediate and self-intermediate scattering functions, F ( k , t ) and Fs(k, t ) , accurate to first order in the volume concentration of particles are constructed, and are solved for suspensions of hard spheres with and without hydrodynamic interaction. For short and long times (semi-) analytic solutions are given; for intermediate times numerical results are presented. The formal correspondence of the limiting values of the time-dependent solutions with the results of Batchelor (1976, 1983) and others for steady sedimentation in polydisperse systems is established.good deal of attention. Table 1 gives a partial list of relevant papers. A detailed explanation of the symbols used is given in $2. The papers fall into two categories. Those influenced by suspension mechanical work on steady transport phenomena in two-phase flow have tended to use exact hydrodynamic data (which is an important ingredient of the problem), but have provided results appropriate only to special limiting cases of the light-scattering problem. Papers in the second category have attempted (more) complete solutions of the problem but have used approximate hydrodynamic data. For many papers in both categories it is not always clear to which regime in wavenumber and time the result applies, nor the extent of its accuracy and validity. In this account we seek to combine the better features of both approaches, to use exact hydrodynamics whenever possible, and to provide both a physical explanation for, and a mathematical proof of, the correspondence between the various results which provide partial solutions t o the full problem. The difference between 'exact ' and 'approximate ' hydrodynamics should be amplified since the term exact is intended to convey more than mere accuracy. The solution of two-and many-body problems in hydrodynamics can in general be achieved only numerically, and in consequence the final results are subject to rounding and cut off errors (e.g. see table 2). On the other hand, the analytic expressions given are exact, and can exhibit qualitative differences from results obtained using 'approximate ' data (see for example the discussion of $3.2). In $2 we delineate the various timescales of interest, and in $3 provide a full mathematical formulation for the dilute limit. In $54 and 5 we solve the equations ...