Quality of service (QoS) support is a key attribute for multimedia traffic including video, voice, and data in wireless local area networks (LANs) but is limited in 802.11-based wireless LANs. A polling-based scheme called the point coordination function (PCF) was developed for 802.11 LANs to support the transmission of multimedia traffic. However, the PCF is not able to meet the desired practical traffic differentiation requirements for real-time data. This paper describes a QoS support polling scheme based on the IEEE 802.11 medium access control (MAC) protocol. The scheme uses a two-level polling mechanism with the QoS classes differentiated by two different access policies. Stations with higher priority traffic such as key or real-time data form the first level and can access the common channel through an exhaustive access policy. Other stations with lower priority traffic form the second level and can access the channel through a gated access policy. A system model based on imbedded Markov chain theory and a generation function were setup to explicitly analyze the mean information packet waiting time of the two-level polling scheme. Theoretical and simulation results show that the new scheme efficiently differentiates services to guarantee better QoS and system stability. Quality of service (QoS) support is a key attribute for multimedia traffic including video, voice, and data in wireless local area networks (LANs) but is limited in 802.11-based wireless LANs. A polling-based scheme called the point coordination function ( PCF) was developed for 802.11 LANs to support the transmission of multimedia traffic. However, the PCF is not able to meet the desired practical traffic differentiation requirements for real-time data. This paper describes a QoS support polling scheme based on the IEEE 802.11 medium access control (MAC) protocol. The scheme uses a two-level polling mechanism with the QoS classes differentiated by two different access policies. Stations with higher priority traffic such as key or real-time data form the first level and can access the common channel through an exhaustive access policy. Other stations with lower priority traffic form the second level and can access the channel through a gated access policy. A system model based on imbedd ed Markov chain theory and a generation function were setup to explicitly analyze the mean information packet waiting time of the two-level polling scheme. Theoretical and simulation results show that the new scheme could differentiates services to guarantee better QoS and system stability.