Abstract This version extends the computation of the liquid-scintillation counting efficiency to electron-capture radionuclides of 30 ⩽ Z ⩽ 54 . The simplified deterministic models of previous versions are replaced by a complete stochastic model, which considers all possible subshells involved in the atomic rearrangement of the atom. The program can simulate samples in the gel phase, including the effects of the micelles on the counting efficiency. These effects have been found to be useful for building nanodosimeters based on gel scintillators. Program summary Title of program: MICELLE Catalogue identifier: ACPU_v3_0 Program summary URL: http://cpc.cs.qub.ac.uk/summaries/ACPU_v3_0 Program obtainable from: CPC Program Library, Queen's University of Belfast, N. Ireland Licensing previsions: none Computers revisions: any IBM PC compatible with 80386 or higher Intel processors Operating systems under which the program has been tested: MS-DOS and higher systems Programming language used: FORTRAN 77 Memory required to execute with typical data: 235 kword No. of bits in a word: 16 No. of lines in distributed program, including test data, etc.: 16 653 No. of bytes in distributed program, including test data, etc.: 358 166 Distribution format: tar.gz Nature of the physical problem: Both β and electron-capture are decay processes characterized by a large variability in energy. In the first case, one single β -particle is emitted per decay following the Fermi distribution. In the second, several electrons (Auger and/or Coster–Kronig) of very different energies can be ejected simultaneously. The detailed simulation of these two electron release processes has practical interest in two situations: (1) to standardize radionuclides with a liquid-scintillation counter, (2) to compute the absorbed dose in the surroundings of a radiolabeled molecule. Method of solution: Although the application of simplified deterministic models is sufficiently accurate for pure β -ray emitters, the large stochastic variability of both electron-capture and internal conversion processes restricts the accuracy of the deterministic models KLM, KLMN and KL 1 L 2 L 3 M to nuclides of low atomic numbers. To extend the applicability of the method to larger nuclei, both M i - and N j -subshells must be included into the model. However, the addition of these outer atom subshells to the deterministic model involves a huge number of atomic rearrangement pathways, requiring from simplifications which are frequently limited to certain nuclides. A more feasible method considers using random numbers to simulate step by step the rearrangement of the atom. Restrictions on the complexity of the problem: The program is restricted to radionuclides of atomic numbers within the interval 30 ⩽ Z ⩽ 54 . This version ignores the photoionization quench correction, which can be obviated for Z ⩾ 30 . On the other hand, the simulation of the mechanisms of multiple ionization require from more elaborated models for Z > 54 . Experiments with gases are only available for nuclides with atomic numbers larger than that of 131 I, for which the emission of Auger electrons, and consequently the ionization of xenon ( Z = 54 ), stops for transitions outer than N 4 O 2 O 2 .