OBJECTIVES: Drug prescriptions recorded in healthcare administrative databases (HADs) can be an indicator of moderate health outcomes (undetectable through hospitalization or death registries) correlated to particulate matter (PM) exposure. This study is aimed at assessing the short-term effect of PM10 exposure on the risk of cardiorespiratory drug prescription in Lombardy, a region in northern Italy.  METHODS: For each resident in seven cities of Lombardy, we identified all prescriptions of selected respiratory, cardiovascular and antidiabetic treatments recorded during 2005-2006 in data warehouse DENALI, which gathers HADs of Lombardy health system. The Regional Environmental Protection Agency of Lombardy provided time-series of daily mean PM10 concentration. We applied a time-stratified case-crossover design matched by day of week and fitted separate Cox proportional hazard models for each respiratory and cardiovascular treatment. Confounding was accounted for using a method, which we developed, based on the time-series of prescription of antidiabetics. Analyses were replicated for delayed effects of PM up to 6 days and for warm and cold season. RESULTS: The study area counted 470,300 residents, requiring 655,805 prescriptions. Mean PM10 concentration was 48µg/m3 (SD 32µg/m3). Overall, we estimated that rises in PM10 concentration were associated with an immediate increment in the risk of prescription of inhalant adrenergics (0.32% for increments of 10µg/m3 in PM10 concentration; 95%CI:0.00,0.65), antiarrhythmics (0.52%; 95%CI:0.16,0.87) and nitrates (0.51%; 95%CI:0.27,0.76). Increased PM10 exposure was also positively associated with the prescription of inhalant glucocorticoids during the warm season and of inhalant adrenergics, antiarrhythmics and nitrates during the cold season. CONCLUSIONS: The significant association that we detected between PM concentration and drug prescriptions suggests that PM exposure may impact public health not only through severe but also through moderate adverse events. Further investigation is needed and given the usual difficulty retrieving information on moderate outcomes, HADs represent a valuable data source.