Several mobility models are available for simulating WLAN, with different impacts on network performance. This work deals with the impact of the assumed mobility model on two key teletraffic variables involved in the planning of the network: the cell residence time (i.e., time connected to an access point) and the handoff rate. These two variables are studied in different scenarios for WLANs designed for pedestrians. For this purpose, discrete event simulations are run with different mobility patterns and number of access points. The time between changes of access point (i.e., handoffs) is studied as a random variable. This research proves the importance of correctly selecting the assumed mobility pattern, as it has a strong impact on the number of handoffs. The probability density function of the cell residence time is also studied as a combination of a distribution that models fast disassociation events (i.e., short ping-pongs between two access points) and a gamma or lognormal distribution, depending on the mobility pattern, which model longer dwells.