Abstract: Using meteorological observations, upper-air observations and numerical weather prediction products from 2008 to 2012 in Lhasa city, 68 meteorological factors associated with air pollutants were picked up and standardized to establish the regression equations of SO₂, NO₂ and PM₁₀ in different seasons. The human activities such as Incense and fireworks in religious activities and festivals holidays were revised forecast for the results obtained concentration of pollutants. The data of 2013 were used to test forecast and took a good prediction results. In spring, SO₂’s average absolute error (AAE) is between 0.001 and 0.0022 mg/m³, average relative error (ARE) lays 15.3%-27.0%; NO₂’s AAE is between 0.003 and 0.005 mg/m³, ARE lays 17.2%-27.5%; PM₁₀’s AAE is between 0.006 and 0.019 mg/m³, ARE lays 16.6%-28.6%.The Air Pollution Index(API) forecast accuracy is between 78.2%-87.8%. The prediction method can fulfill the needs of air pollutant concentrations forecast service in Lhasa city.