A model has been developed for predicting median levels of traffic noise in a variety of local conditions. Sound sources are grouped into two categories: major thoroughfares with average daily traffic volume greater than 5000 vehicles are considered individually as line sources with uniform distribution of traffic and all other streets are grouped into a set of square areas. The model makes use of readily available traffic and census tract data. It takes into account interruptions in the traffic stream. Noise levels predicted by the model are in agreement with levels observed during a noise study in Cincinnati, Ohio. The correlation coefficient was 0.90, and 64% of the predicted levels fell within ±2.5 dB of the observed levels. Both sections of the model, involving line and area sources, performed equally well and no systematic bias was observed. This simple model is proposed as a practical working tool to predict noise levels under any existing or proposed set of conditions.