Recent epidemiological and toxicological studies have shown associations between particulate matter and human health. However, the estimates of adverse health effects are inconsistent across many countries and areas. The stratification and interaction models were employed within the context of the generalized additive Poisson regression equation to examine the acute effects of fine particles on respiratory health and to explore the possible joint modification of temperature, humidity, and season in Beijing, China, for the period 2004-2009. The results revealed that the respiratory health damage threshold of the PM2.5 concentration was mainly within the range of 20-60 μg/m(3), and the adverse effect of excessively high PM2.5 concentration maintained a stable level. In the most serious case, an increase of 10 μg/m(3) PM2.5 results in an elevation of 4.60 % (95 % CI 3.84-4.60 %) and 4.48 % (95 % CI 3.53-5.41 %) with a lag of 3 days, values far higher than the average level of 0.69 % (95 % CI 0.54-0.85 %) and 1.32 % (95 % CI 1.02-1.61 %) for respiratory mortality and morbidity, respectively. There were strong seasonal patterns of adverse effects with the seasonal variation of temperature and humidity. The growth rates of respiratory mortality and morbidity were highest in winter. And, they increased 1.4 and 1.8 times in winter, greater than in the full year as PM2.5 increased 10 μg/m(3).