Dynamic light scattering has become a standard technique for investigating colloidal suspensions and polymer solutions. The experimental field autocorrelation function (g) over cap(1)$(t) can often be well modelled by a Laplace transform relating (g) over cap(1)$(t) to a distribution of decay times A(tau). In simple systems A(tau) can usually be directly related to a distribution of molecular weights, particle sizes, diffusion coefficients or other physically relevant quantities. With constrained regularization methods, the parameter-free estimation of A(tau) has become straightforward. In complex systems, the resulting A(tau) may contain several components the identification of which is not always obvious. The problem often originates in a superposition of diffusive and angle-independent components that have different variations of their respective decay times with the scattering vector. A method is presented based on a simultaneous fit of several autocorrelation functions measured at several different scattering angles, which, using simple and reasonable assumptions, yields a robust analysis of the spectra of decay times. The application of the method is illustrated on simulated autocorrelation functions and also on real experimental data obtained on a variety of different polymer systems.