Spherical radioactive caesium (cs)-bearing microparticles (csMps) were emitted during the fukushima Dai-ichi Nuclear Power Plant (FDNPP) accident in March, 2011. The emission source (timing) and formation process of these particles remain unclear. In this study, the isotopic ratios of uranium (235 U and 238 U) and caesium (133 Cs, 134 Cs, 135 Cs, and 137 Cs) isotopes in the five spherical CsMPs (ca. 2 μm in size) sampled at 50 km west of the FDNPP were determined using secondary ion mass spectrometry and laser ablation-ICPMS, respectively. Results showed that the 235 U/ 238 U ratios of csMps were homogeneous (1.93 ± 0.03, N = 4) and close to those estimated for the fuel cores in units 2 and 3, and that the Cs isotopic ratios of CsMP were identical to those of units 2 and 3. These results indicated that U and Cs in the spherical CsMPs originated exclusively from the fuel melt in the reactors. Based on a thorough review of literatures related to the detailed atmospheric releases of radionuclides, the flow of plumes from the fDnpp reactor units during the accident and the U and cs isotopic ratio results in this study, we hereby suggest that the spherical CsMPs originate only from the fuel in unit 2 on the night of 14 March to the morning of 15 March. The variation range of the analysed 235 U/ 238 U isotopic ratios for the four spherical particles was extremely narrow. Thus, U may have been homogenised in the source through the formation of fuel melt, which ultimately evaporating and taken into CsMPs in the reactor and was released from the unit 2. On 11 March 2011, the Great East Japan Earthquake triggered the Fukushima Dai-ichi Nuclear Power Plant (FDNPP) accident. Radioactive materials were released to the atmosphere and have caused various environmental problems 1-4. Among the six nuclear reactors in the FDNPP, the reactors of units 1-3 were in operation at the time of the earthquake 5. In the accident, hydrogen explosions occurred in the nuclear reactor buildings of units 1, 3 and 4 (at 15:36 JST on 12 March, 11:01 JST on 14 March and 6:14 JST on 15 March, respectively), but the explosion in unit 4 occurred in the reverse flow of hydrogen generated in unit 3 6. Unit 4 had been shut down from 2010, and all of the fuel were kept in the spent fuel pool (SFP), suggesting that no radionuclide was emitted from unit 4, whereas units 1 and 3 can be the sources of the radionuclides. Meanwhile, the release of radionuclides from reactor pressure vessel (RPV) or blowout panel in unit 2 was also an important source of the radionuclides 7 .