As feature sizes continue to shrink, traditional corner analysis, because of its limitations and pessimism [1], is being gradually supplanted by Statistical Static Timing Analysis (SSTA) as our major timing tool. Recent works [2-4] propose non-linear models handling Gaussian and non-Gaussian variation sources, which is a great progress relative to linear dependency on Gaussian process parameters presented in [5-6]. The authors of [7] propose a simple and practical approach of SSTA. After having analyzed impacts of process variations on delay and probable phenomena of circuit, the authors simplify the cells timing characterization, and the basic operations MAX/MIN for random variables. Following the thought of simplicity, we propose in this paper a path-based approach to propagate probability density function (PDF) with the help of conditional moments (conditional mean and conditional variance). With these two probability concepts, plus a new analytical model of input signal and the employment of active load, path delay distribution can be estimated taking input slope and output load variations into account. What's more, we attempt to tackle the problem never been mentioned: estimate of structure correlations, which come from the fact that output signal of one cell is input signal of the next stage