Aims: Distorted wall shear stress (WSS) in patients with type 2 diabetes mellitus (T2DM) may be partly explained by an altered red blood cell aggregation tendency (RAT) on viscosity at low shear rate (SR). The present study evaluates viscosity modeling by implementation of hematocrit and RAT in patients with and without T2DM (non-T2DM). Methods: A Couette viscometer and LORCA aggregometer provided viscosity and RAT on 6 shear rates in 55 patients (46-78 yrs, 66% male, T2DM: n = 28), following informed consent. Using a K-fold cross-validation, two linear mixed models predicted by SR and Hct and by SR, Hct and RAT were compared. Results: In non-T2DM modeling was improved in relatively low RATs (48%, p = 1.0 × 10-11) and became worse in relatively high RATs (-18%, p = 0.019). In T2DM the opposite was observed, as modeling became worse in relatively low RATs (-16%, p = 0.001) but was improved in relatively high RATs (22%, p = 0.022). Conclusions: In addition to confirming previous research, major differences in modeling improvement between T2DM and non-T2DM were found. Especially patients with T2DM, a high RAT and often high viscosity at low SR benefit from a more accurate viscosity modeling. Further studies should evaluate how these findings affect WSS in these patients.