Data evaluation is a crucial step when it comes to the determination of accurate and precise isotope ratios computed from transient signals measured by multi-collector–inductively coupled plasma mass spectrometry (MC-ICPMS) coupled to, for example, laser ablation (LA). In the present study, the applicability of different data evaluation strategies (i.e. ‘point-by-point’, ‘integration’ and ‘linear regression slope’ method) for the computation of 235U/238U isotope ratios measured in single particles by LA-MC-ICPMS was investigated. The analyzed uranium oxide particles (i.e. 9073-01-B, CRM U010 and NUSIMEP-7 test samples), having sizes down to the sub-micrometre range, are certified with respect to their 235U/238U isotopic signature, which enabled evaluation of the applied strategies with respect to precision and accuracy. The different strategies were also compared with respect to their expanded uncertainties. Even though the ‘point-by-point’ method proved to be superior, the other methods are advantageous, as they take weighted signal intensities into account. For the first time, the use of a ‘finite mixture model’ is presented for the determination of an unknown number of different U isotopic compositions of single particles present on the same planchet. The model uses an algorithm that determines the number of isotopic signatures by attributing individual data points to computed clusters. The 235U/238U isotope ratios are then determined by means of the slopes of linear regressions estimated for each cluster. The model was successfully applied for the accurate determination of different 235U/238U isotope ratios of particles deposited on the NUSIMEP-7 test samples.Electronic supplementary materialThe online version of this article (doi:10.1007/s00216-012-6674-3) contains supplementary material, which is available to authorized users.