The standard metrics for integrated circuits' analysis and characterization in production processes usually assume that the process under investigation is characterized by a normal distribution. However, the data met in practice are not always normal and the yield estimates may be inaccurate. In this paper we propose estimating the yield by using a distribution fitting flow. The selected distribution types proved to estimate more accurate yields, with lower variance of the estimates. The distribution models proved to be a reliable tool also for integrated circuits' characterization in terms of specification limits' determination.