Two models for predicting in-city road-traffic noise pollution of Mashhad have been obtained. Traffic volume, composition, and speed have been chosen as model's parameters. Vehicles were classified into light cars and medium and heavy trucks. Reference emission level of each group was determined experimentally based on perpendicular propagation from central lane of traffic road. Simultaneous measurements of noise level and vehicle flow and composition were done. Two mathematical models have been proposed by the use of genetic algorithms which can be used for calculating L_eq. These models have been validated against noise data. Subsequently, measured traffic noise has been compared with calculated ones, using developed models, and a relatively good agreement has been obtained among them. The models are found accurate within ±1% and can be used for flat road noise prediction.