Human biomonitoring comprises the determination of biomarkers in body-fluids, cells and tissues. Biomarkers are generally assigned to one of three classes, namely, biomarkers of exposure, effect or susceptibility. Since biomarkers represent steps in an exposure-disease continuum, their application in epidemiological studies ('molecular epidemiology') shows promise. However, to be a predictor of disease, a biomarker has to be validated. Validation criteria for a biomarker include intrinsic qualities such as specificity, sensitivity, knowledge of background in the population, existence of dose-response relationships, degree of inter- and intra-individual variability, knowledge of the kinetics, confounding and modifying factors. In addition, properties of the sampling and analytical procedures are of relevance, including constraints and non-invasiveness of sampling, stability of sample as well as simplicity, high sensitivity, specificity and speed of the analytical method. It is of particular importance to prove by suitable studies that the biomarker of exposure indicates the actual exposure, the biomarker of effect strongly predicts the actual risk of disease and the biomarker of susceptibility actually modifies the risk. Biomonitoring of the exposure to complex mixtures such as polluted ambient air, diesel exhaust or tobacco smoke is a particular challenge since these exposures have many constituents in common and many people were exposed to more than one of these mixtures. Data on the exposure to polycyclic aromatic hydrocarbons (PAH) and benzene from ambient air, diesel exhaust and tobacco smoke will be presented. In addition, some source-specific biomarkers such as nitro-arenes and nicotine metabolites as well as their application in population groups will be discussed. The second part of the presentation addresses the application of biomarkers for assessing so called 'potentially reduced exposure products' (PREPs). According to a recent report of the Institute of Medicine (USA), "reducing risk of disease by reducing exposure to tobacco toxicants is feasible" and "surrogate biological markers that are associated with tobacco-related diseases could be used to offer guidance as to whether or not PREPs are likely to be risk-reducing." In general, the same validation criteria apply as discussed above. In addition, it is suggested that a panel of biomarkers should be used, representing both smoke phases (gas and particulate phase) and the various chemical classes of smoke constituents (e.g., carbonyls, benzene, PAH, tobacco-specific nitrosamines, aromatic amines). Also, a panel of biomarkers of effect should cover the major known adverse effects of smoking (e.g., oxidative stress, inflammatory processes, lipid peroxidation, lipometabolic disorders, mutagenic effects). Biomarkers of nicotine and carbon monoxide uptake are of interest for evaluating the smoking and inhalation behavior, respectively. Finally, suitable study designs for evaluating PREPs are discussed. It is concluded that suitable biomarkers for assessing the exposure to complex mixtures such as ambient air, diesel exhaust and tobacco smoke as well as for evaluating the exposure-reducing properties of PREPs are already available. Future efforts should focus on the development and validation of biomarkers of effect.