While fading effects have long been combatted in 2G wireless networks, primarily devoted to voice calls, they are now seen as an opportunity to increase the capacity of 3G networks that incorporate data traffic. The packet delay tolerance of data applications, such as file transfers and Web browsing for instance, allows the system flexibility in scheduling a user's packets. So-called opportunistic scheduling, which ensures transmission occurs when radio conditions are most favourable, is the key component of currently developed systems like cdma2000 IS-856 and UMTS HSDPA. We compare the performance of some scheduling schemes using a flow-level approach where the random evolution of the number of ongoing flows is explicitly taken into account.