Inhalation exposure to uranium aerosols can be a concern in nuclear fuel fabrication. The ICRP provides default absorption parameters for various uranium compounds but also recommends determination of material-specific absorption parameters to improve dose calculations for individuals exposed to airborne radioactivity. Aerosol particle size influences internal dosimetry calculations in two potentially significant ways: the efficiency of particle deposition in the various regions of the respiratory tract is dependent on aerodynamic particle size, and the dissolution rate of deposited materials can vary according to particle size, shape, and porosity because smaller particles tend to have higher surface-to-volume ratios than larger particles. However, the ICRP model assumes that deposited particles of a given material dissolve at the same rate regardless of size and that uptake to blood of dissolved material normally occurs instantaneously in all parts of the lung (except the anterior portion of the nasal region, where zero absorption is assumed). In the present work, the effect of particle size on dissolution in simulated lung fluid was studied for uranium aerosols collected at the plant, and its influence on internal dosimetry calculations was evaluated. Size fractionated uranium aerosols were sampled at a nuclear fuel fabrication plant using portable cascade impactors. Absorption parameters, describing dissolution of material according to the ICRP Human Respiratory Tract Model, were determined in vitro for different size fractions using simulated lung fluid. Samples were collected at 16 time-points over a 100-d period. Uranium content of samples was determined using inductively coupled plasma mass spectrometry and alpha spectrometry. In addition, supplementary experiments to study the effect of pH drift and uranium adsorption on filter holders were conducted as they could potentially influence the derived absorption parameters. The undissolved fraction over time was observed to vary with impaction stage cut-point at the four main workshops at the plant. A larger fraction of the particle activity tended to dissolve for small cut-points, but exceptions were noted. Absorption parameters (rapid fraction, rapid rate, and slow rate), derived from the undissolved fraction over time, were generally in fair agreement with the ICRP default recommendations for uranium compounds. Differences in absorption parameters were noted across the four main workshops at the plant (i.e., where the aerosol characteristics are expected to vary). The pelletizing workshop was associated with the most insoluble material and the conversion workshop with the most soluble material. The correlation between derived lung absorption parameters and aerodynamic particle size (impactor stage cut-point) was weak. For example, the mean absorption parameters derived from impaction stages with low (taken to be <5 μm) and large (≥5 μm) cut-points did not differ significantly. Drift of pH and adsorption on filter holders appeared to be of secondary importance, but it was found that particle leakage can occur. Undissolved fractions and to some degree derived lung absorption parameters were observed to vary depending on the aerodynamic size fraction studied, suggesting that size fractionation (e.g., using cascade impactors) is appropriate prior to conducting in vitro dissolution rate experiments. The 0.01-0.02 μm and 1-2 μm size ranges are of particular interest as they correspond to alveolar deposition maxima in the Human Respiratory Tract Model (HRTM). In the present work, however, the dependency on aerodynamic size appeared to be of minor importance, but it cannot be ruled out that particle bounce obscured the results for late impaction stages. In addition, it was noted that the time over which simulated lung fluid samples are collected (100 d in our case) influences the curve-fitting procedure used to determine the lung absorption parameters, in particular the slow rate that increased if fewer samples were considered.