The influence of temperature on the stability of human serum albumin (HSA) pharmaceutical preparations has been studied by size-exclusion high-performance liquid chromatography with multi-angle laser-light-scattering detection and by particle-size analysis. The behaviour of HSA in two pharmaceutical preparations stored at different temperatures (40, 55 and 70°C) followed the same pattern—an increase in the relative percentage of dimer (MW 132 000 Da) and aggregate (MW > 200 000 Da), and then a decrease in the concentration of all species and, finally, sudden protein coagulation. These results suggest a time- and temperature-dependent process. At 70°C, monomer only was detected for both preparations; the amount remaining was 83 and 72% for formulations A and B, respectively. Analysis of size-distribution curves also seems to confirm these results. Initially, three distributions were observed with length-volume mean diameters (dl,v)of 1.67, 10.6 and 57 μm. After 80 days at 55°C, only two distributions were observed, with dl,v of 3.07 and 76 μm. An additional study using pure HSA at different concentrations (0.3, 2.5, 5 and 10% w/v) and storage at 75°C was performed to determine the influence of the concentration of auxiliary substances and of the HSA. Only when the HSA concentration was 0.3% w/v did the remaining fraction of HSA fit a Prout-Thompkins nucleation model. Initially three distributions with mean sizes of 2, 20 and 40 μm were observed whereas at the end of the assay only one distribution, mean size 129 μm, was seen. The methodology used enabled us to separate the HSA degradation products and to determine the absolute molecular weight of albumin monomer and dimer. It is possible to conclude that the degradation mechanism for the formulations studied is complex, and that it is possible to fit the degradation data to Prout-Thompkins kinetics only when the concentration of HSA is low enough (0.3% w/v).