Simulated Annealing (SA), a Monte Carlo method, is used to obtain discrete relaxation spectra from small strain oscillatory shear measurements. The SA algorithm does so by evaluating randomly generated trial values of relaxation moduli, g(k), and relaxation times, lambda(k). If the solution of a trial point gives a lower function value, it is accepted as a new minimum. The solution can escape from local minima by using the Metropolis criterion which accepts an uphill move during minimisation given a certain probability. SA searches for the global optima, i.e. the best fit. In order to avoid fitting of noise, only a limited number of modes are fitted. Constraints can easily be imposed on the solution to ensure that the relaxation times are kept within the limits of sampling localisation and to avoid the calculation of negative g(k) and lambda(k). The results demonstrate SA's ability to fit a relaxation spectrum with a short computation time. The quality of the fit is determined by the truncated data range, and, to a certain degree, by a chosen sampling localisation. limit.