Interconnects are an important constituent of any large scale integrated circuit, and accurate interconnect analysis is essential not only for post-layout verification but also for synthesis. For instance, extraction of interconnect capacitance is needed for the prediction of interconnect-induced delay, crosstalk, and other signal distortion related effects that are used to guide IC routing and floor planning. The continuous progress of semiconductor technology is leading ICs to the era of 45 nm technology and beyond. However, this progress has been associated with increasing variability during the manufacturing processes. This variability leads to stochastic variations in geometric and material parameters and has a significant impact on interconnect capacitance. It is therefore important to be able to quantify the effect of such process induced variations on interconnect capacitance. In this thesis, we have worked on a methodology towards modeling of interconnect capacitance in the presence of geometric uncertainties. More specifically, a methodology is proposed for the finite element solution of Laplace's equation for the calculation of the per-unit-length capacitance matrix of a multi-conductor interconnect structure embedded in a multi-layered insulating substrate and in the presence of statistical variation in conductor and substrate geometry. The proposed method is founded on the idea of defining a single, mean geometry, which is subsequently used with a single finite element discretization, to extract the statistics of the interconnect capacitance in an expedient fashion. We demonstrate the accuracy and efficiency of our method through its application to the extraction of capacitances in some representative geometries for IC interconnects.