The component of air quality that is concentration of PM22 s has great significance in our life human health and city management. Each and every activities of living things are impacted by the quality of air worldwide like air quality has its impact on mode of travelling, urban planning, policy makers, government rules and regulations and so on. However, in most of cities contains only limited number of monitoring stations for quality which is not effective as we know that the dynamic nature of air quality which varies from place to place even between neighboring regions. In this paper, a random forest approach for predicting air quality (RAQ) is proposed for urban sensing systems. The data collected from urban sensing includes meteorology data, road information, real-time traffic status and point of interest (POI) distribution. The random forest algorithm is introduced for data training and prediction. The performance analysis of RAQ is evaluated with real-time data from Delhi. When the comparison was done with other three algorithm, this approach achieves better prediction precision. Unique results are observed from the experiments that the air quality can be inferred with amazingly high accuracy from the data which are obtained from urban sensing.