Purpose Small‐scale dosimetry studies generally consider an artificial environment where the tumors are spherical and the radionuclides are homogeneously biodistributed. However, tumor shapes are irregular and radiopharmaceutical biodistributions are heterogeneous, impacting the energy deposition in targeted radionuclide therapy. To bring realism, we developed a dosimetric methodology based on a three‐dimensional in vitro model of follicular lymphoma incubated with rituximab, an anti‐CD20 monoclonal antibody used in the treatment of non‐Hodgkin lymphomas, which might be combined with a radionuclide. The effects of the realistic geometry and biodistribution on the absorbed dose were highlighted by comparison with literature data. Additionally, to illustrate the possibilities of this methodology, the effect of different radionuclides on the absorbed dose distribution delivered to the in vitro tumor were compared. Methods The starting point was a model named multicellular aggregates of lymphoma cells (MALC). Three MALCs of different dimensions and their rituximab biodistribution were considered. Geometry, antibody location and concentration were extracted from selective plane illumination microscopy. Assuming antibody radiolabeling with Auger electron (125I and 111In) and β− particle emitters (177Lu, 131I and 90Y), we simulated energy deposition in MALCs using two Monte Carlo codes: Geant4‐DNA with “CPA100” physics models for Auger electron emitters and Geant4 with “Livermore” physics models for β− particle emitters. Results MALCs had ellipsoid‐like shapes with major radii, r, of ~0.25, ~0.5 and ~1.3 mm. Rituximab was concentrated in the periphery of the MALCs. The absorbed doses delivered by 177Lu, 131I and 90Y in MALCs were compared with literature data for spheres with two types of homogeneous biodistributions (on the surface or throughout the volume). Compared to the MALCs, the mean absorbed doses delivered in spheres with surface biodistributions were between 18% and 38% lower, while with volume biodistribution they were between 15% and 29% higher. Regarding the radionuclides comparison, the relationship between MALC dimensions, rituximab biodistribution and energy released per decay impacted the absorbed doses. Despite releasing less energy, 125I delivered a greater absorbed dose per decay than 111In in the r ~ 0.25 mm MALC (6.78·10−2 vs 6.26·10−2 µGy·Bq−1·s−1). Similarly, the absorbed doses per decay in the r ~ 0.5 mm MALC for 177Lu (2.41·10−2 µGy·Bq−1·s−1) and 131I (2.46·10−2 µGy·Bq−1·s−1) are higher than for 90Y (1.98·10−2 µGy·Bq−1·s−1). Furthermore, radionuclides releasing more energy per decay delivered absorbed dose more uniformly through the MALCs. Finally, when considering the radiopharmaceutical effective half‐life, due to the biological half‐life of rituximab being best matched by the physical half‐life of 177Lu and 131I compared to 90Y, the first two radionuclides delivered higher absorbed doses. Conclusion In the simulated configurations, β− emitters delivered higher and more uniform absorbed dose than Auger electron emitters. When considering radiopharmaceutical half‐lives, 177Lu and 131I delivered absorbed doses higher than 90Y. In view of real irradiation of MALCs, such a work may be useful to select suited radionuclides and to help explain the biological effects.< Réduire