A human biomonitoring (HBM) survey in four areas affected by natural or anthropogenic arsenic pollution was conducted in Italy within the framework of the SEpiAs project. A questionnaire, including the exploration of risk perception (RP) regarding environmental hazards and access to and trust in information, was administered to 282 subjects stratified by area, gender and age. The survey was designed to investigate how populations living in polluted areas could adopt prevention-oriented habits, fostered by the awareness of existing risks and, in addition, how increased knowledge of RP and information flows could support researchers in identifying recommendations, and presenting and disseminating HBM results. This study characterizes the four areas in terms of RP and access to and trust in environmental information, and provides insights into the influence of RP and environmental information on food consumption. For the data analysis, a combined random forest (RF) and logistic regression approach was carried out. RF was applied to the variables derived from the questionnaire in order to identify the most important in terms of the aims defined. Associations were then tested using Fisher's exact test and assessed with logistic regression in order to adjust for confounders. Results showed that the perception of and personal exposure to atmospheric and water pollution, hazardous industries and waste, hazardous material transportation and waste was higher in geographical areas characterized by anthropogenic pollution. Citizens living in industrial areas appeared to be aware of environmental risks and had more confidence in environmental non-governmental organizations (NGOs) than in public authorities. In addition, they reported an insufficient circulation of information. Concerning the influence of RP and environmental information on food consumption, a high perception of personal exposure to atmospheric pollution and hazardous industries was associated with a lower consumption of local fish. In conclusion, different RPs and information flow patterns were observed in areas with arsenic of natural origin or in industrial contexts. These findings may be useful for targeted risk communication plans in support of risk-management strategies. •Risk perception and access to information was investigated in a sample of 282 subjects in Italy.•A combined random Forest and logistic regression approach was applied.•Risk perception is higher in geographical areas characterized by anthropogenic pollution.•Citizens living in industrial areas are more confident in environmental NGOs than in public authorities.