Odour nuisance has a not negligible impact on the quality of life of the population affected by it, yet tracing the source of malodorous releases is challenging. Numerical models may be profitably used, to support both the identification of the potential source areas and the execution of timely field measurements. In this frame, a new methodology has been developed with the aim of tracing backward the origin of odour nuisance, starting from citizens’ notifications. A numerical procedure processing the alerts received through the Web-App NOSE, has been developed to produce the proper input to the dispersion module of the SMART meteo-dispersive modelling system. Then, a post-processing numerical code elaborates the simulation output based on a combination of puffs of Lagrangian stochastic particles, released backward in time from the locations and at the time of the citizens’ notifications. This novel approach is presented, it is applied to three real case studies occurred in the area of Siracusa province (Sicily, Italy), then related results are discussed. The new methodology proved to be reliable and effective in identifying the most probable zones that can originate the nuisance. Display omitted •Citizen cooperation in support to odour nuisance tracing.•Lagrangian backward numerical modelling to identify source areas.•A new approach to use citizens’ notifications as input to numerical model simulations.•A new methodology to elaborate probability distribution maps of source locations.