Forest fires are a major emission source of pollutants to the atmosphere with several adverse impacts on human health and ecosystems either at local, regional or global scales. Entire populations can be exposed to hazardous concentrations of toxins, especially when fire occurs in the vicinity of cities, as recently drawn by large wildfires in Southeast Asia, Australia, South America or Russia, but also Southern Europe, namely in Portugal. Nowadays there is a growing concern about the increase in both the frequency and the severity of forest fires spreading in the urban wildland interface (WUI) as a consequence of the escalation of urban development among or adjacent to wildlands. The main purpose of this paper is to estimate the impact of wildland forest fires on the air quality of a city in Portugal. Aiming to take into account the effect of a larger atmospheric scale in the very local urban one, a multi-scale approach was adopted, which implied the link between a chemistry-transport mesoscale model (LOTOS-EUROS) and a microscale computational fluid dynamics (CFD) model. Forest fire emissions were estimated based on specific southern European emission factors, type of vegetation, area burned and fire behaviour, and were incorporated in the emission input data of the numerical modelling system. Results confirm the strong impact of forest fires on the urban air pollution levels. Statistical indicators were used to validate the modelling application through the comparison of results to measured air quality data. This modelling approach has got very good performance skills showing the possibility of applying this kind of system to analyse the relation between forest fires and urban air pollution.