The use of the downhill simplex technique is presented for the deconvolution of particle-size distribution functions from optical-scattering and optical-extinction measurements. Criteria are developed for the assessment of the goodness of fit, and the stabilities of the simplex and the linear-inversion methods are compared. Monte Carlo simulations of synthetic data are used with various random-error levels to demonstrate and to compare the deconvolution results of log-normal distributions from scattering and extinction measurements.