AbstractLong-range simulation of the dispersion of air pollutants in the atmosphere is one of the most challenging tasks in geosciences. Application of precise and fast numerical models in risk management and decision support can save human lives and can diminish consequences of an accidental release. Disaster at Fukushima Daiichi nuclear power plant has been the most serious event in the nuclear technology and industry in the recent years. We present and discuss the results of the numerical simulations on dispersion of Fukushima-derived particulate ¹³¹I and ¹³⁷Cs using a global scale Lagrangian particle model. We compare concentrations and arrival times, using two emission scenarios, with the measured data obtained from 182 monitoring stations located all over the Northern Hemisphere. We also investigate the homogenization of isotopes in the atmosphere. Peak concentrations were predicted with typical accuracy of one order of magnitude showing a general underestimation in the case of ¹³¹I but not for ¹³⁷Cs. Tropical and Arctic plumes, as well as the early detections in American and European midlatitudes were generally well predicted, however, the later regional-scale mixing could not be captured by the model. Our investigation highlights the importance of the parameterization of free atmospheric turbulence.