In the token ring network different types of services gives different delay characteristics. Results to date have shown that in asymmetric traffic, exhaustive service gives more delay to lightly loaded stations where ordinary service wastes time in circulating the token after each transmission. In general, there is a need for more efficient service, which is compromised between ordinary and exhaustive service. Ordinary and exhaustive service are analyzed in this thesis, and a new service, adaptive service, is proposed. By using timer and counter, adaptive service dynamically changes token holding time at the station. Different types of delay characteristics are derived from their respective simulation models. The results indicate that proposed adaptive service has superior delay characteristics when compared with ordinary and adaptive service in asymmetric and symmetric traffic.