In this paper we assessed effects of changes of meteorological drivers, taken from datasets of observational records and modelling outputs, and human-made pollution, derived from records of energy production, on the mainly wintertime extreme observed values of urban particulate matter (PM) concentrations in the relative vicinity of coal fuelled thermoelectric power plants (TPPs) in Montenegro and Serbia. We used wavelet transform analysis, together with the dependency analysis and analysis of averages of climatic conditions, to study temporal dynamics of urban air pollution extremes in the vicinity of TPPs, the coincidence of their changes with observed levels of SO2 and NO2 concentrations in the air, and dependence of PM changes on several possible meteorological and anthropogenic drivers. We found that PM variations in urban areas are most probably caused by PM-SO2/NO2 coincidences that appear after a 2- to 3-h time lags needed for transformation of SO2/NO2 TPP emissions into PM particles, if pollution is caused by TPP emissions alone. When other causes of PM variations than the TPP production exist, we found that PM-SO2/NO2 correlations appear at time ranges from several hours to several days. In our analysis only the changes in the planetary boundary layer height (PBLH) coincided with the drive to extremes in PM values, at PBLH levels lower than 300m. Following these findings, we suggested that PM extremes in our sample could be viewed as preconditioned compound events, where TPP and urban heating emissions provide preconditions for PM extremes and PBLH serves as a major meteorological driver to such events.