Statistical analysis is being used widely to evaluate the high-speed link performance for accurate system-level optimization including various device jitter components. One of the key assumptions used in the current methodologies is that data patterns across the bus are assumed to be independent. However, this assumption is not valid anymore when the data bus is coded, invalidating existing statistical approaches. In this paper, the current statistical approach is extended to handle coded cases. Excellent correlation is achieved between our statistical method and transient simulation data. Various examples of the link level performance evaluation are also presented in this paper.