Efficient methods for analyzing vertical handoffs for IP based data sessions are essential for emerging heterogeneous data networks. This is mainly due to the high frequency of vertical handoffs experienced by roaming users in such internetworked environments. This paper presents an analytical approach for evaluating vertical session handoffs in such an environment where the packet arrivals follow a Pareto distribution. The reason behind this assumption is due to the fact that probability distributions with long tails have proven to be better suited for modeling packet inter-arrival times for Internet based data traffic. The analysis and evaluation are applied for a framework previously designed by the authors' for interworking between heterogeneous data networks. Finally, the results obtained from this analysis are compared against the results obtained from a classical queuing model where Poisson arrivals and exponential service times are assumed.