This paper presents a statistical perspective on the relationship of the reliability of a nano-circuit (measured by its probability of failure) and the circuit parameters. The study was motivated by our observation that the reliability estimates using (a commonly used set of) mathematical equations deviated appreciably from the results of a Bayesian Network based reliability calculation software, especially as the circuit size increased. (For comparative purposes, we had utilized the data from thousands of randomly generated circuits.) In this research, we have investigated circuit reliability's sensitivity to circuit parameters besides the ones used in the mathematical equations. For this purpose, we formulated several hypotheses regarding circuit parameters and the reliability, and then used statistical metrics to test them. The findings from this analysis are expected to help create higher accuracy mathematical models for predicting circuit reliability.