The quality indicators of the data and service traffic in multiservice access networks (MSAN) are considered, and traffic forecasting based on the use of neural technology is analyzed. Traffic of multiservice access networks is examined the considering the self-similarity property using fuzzy logic, methods for managing the channel resource and monitoring the onset congestion in the network, models for predicting the waiting time of service traffic, as well as numerical results and interpretation. An approach has been suggested and developed for creating an analytical model for predicting the waiting time of service traffic packet flows in the queue. Analytical expressions are obtained for estimating the channel resource, the average absolute error and variance of the prediction error, and indicators of waiting time in the queue using the apparatus of neural technology.