This thesis reports from an investigation of the adequacy of performance Petri nets (PPNs) for modelling communication protocols and evaluating their performance. The report describes PPNs and their tools. It shows how an abstract specification of an existing communication protocol, the Trivial File Transfer Protocol (TFTP), can be modeled and how implemenmtation details can be added while preserving correctness. We use the tools to predict the performance of TFTP. The performance parameters of the model are obtained from an existing UNIX implementation of TFTP. The predicted performance is compared with measured performance of TFTP on a Sun workstation. From the results we evaluate PPNs with respect to modelling limitations, complexity, expressivity and accuracy.