Monte Carlo optimization with constraints was used to estimate parameters of "skew-normal" selectivity curves for gillnets. Non-linear functions of mesh size were explored in addition to the linear ones described by Wulff (1986). This methodology can be used to estimate the parameters for any distribution used to describe selectivity curves. Better estimates can be obtained than with the search-all method of WuL, which is limited by the amount of computer-running-time required for models with more than four or five parameters. Furthermore, models with parameters defined as polynomial functions of mesh size gave better fits than the simpler linear function models in two of the four cases examined herein.