We used a case-crossover approach to assess the association between air pollution and daily mortality in Shanghai from June 2000 to December 2001. By design, this method can avoid some common concerns about the time-series approach, which was most frequently used to assess the short-term effects of air pollution. Different control periods (unidirectional and bi-directional control samplings) were used for the analysis. With a bi-directional six control sampling approach, the results from a conditional logistic regression model controlling for weather conditions showed that each 10 microg/m3 increase over a 48-h moving average of PM10, SO2 and NO2 corresponds to 1.003 (95%CI 1.001-1.005), 1.016 (95%CI 1.011-1.021), and 1.020 (95%CI 1.012-1.027) relative risk of non-accident mortality, respectively. The association between air pollution and mortality for chronic obstructive pulmonary diseases (COPD) and cardiovascular causes was found to be closer than that for all causes. The results confirmed the deleterious role of the current air pollution level on human health in Shanghai, and provided information on the applicability of case-crossover design in studying the acute health effects of air pollution.