Background: Air pollution, which represents a major environmental risk to human health, comprises a complex mixture of compounds where only little is known about its specific toxicities. Objectives: This study examined the specific toxicities associated with ambient air pollutant mixtures with respect to gas/particle partitioning, particulate matter (PM) size, pollutant polarity and bioaccessibility from PM, and evaluated the contribution of PAHs and their oxygenated and nitrated derivatives (OPAHs, NPAHs). Methods: Air samples (gas phase, PM10 and size- segregated PM), were collected at urban (in winter and summer) and background (winter) sites in the Czech Republic. The total and bioaccessible concentrations were addressed using organic solvent extraction and simulated lung fluid extraction, respectively. Organic extracts were also further fractionated according to polarity. Aryl hydrocarbon receptor (AhR)-mediated activity, anti-/estro- genicity, anti-/androgenicity, thyroid receptor (TR)-mediated activity and cytotoxicity for bronchial cells were determined by human cell-based in vitro bioassays. The contribution of studied compounds to observed effects was assessed by both modelling and reconstructing the mixtures. Results: Significant effects were detected in the sub-micrometre size fraction of PM (estrogenicity, androgenicity, TR- and AhR-mediated activities) and in the gas phase (TR-mediated activity, antiandrogenicity). Compounds interacting with TR showed high bioaccessibility to simulated lung fluid. Relatively lower bioaccessibility was observed for estrogenicity and AhR-mediated activity. However, the toxicity testing of reconstructed mixtures revealed that the targeted pollutants are not the main contributors, except for urban PM air pollution in winter, where they accounted for 5-88% of several effects detected in the original complex environmental samples. Discussion: Studied toxicities were mostly driven by polar compounds largely attributed to the easily inhalable PM1, which is of high relevance for human health risk assessment. Except of parent PAHs in some cases, the targeted compounds contributed to the detected effects mostly to a relatively low extent implying huge data gaps in terms of endocrine disruptive potencies of targeted substances and the significance of other polar compounds present in ambient air.