Background: The setting of normovolemic anemia is required for a variety of research applications, such as testing of novel medication for anemia treatment. Unfortunately, large animal models using full blood draw and replenishment with balanced electrolyte solution (BES) lead to bleeding complications, as coagulation factors and platelets are also drawn. We therefore aimed to establish a model of selective red blood cell (RBC) depletion to the main endpoint of hemoglobin (Hgb) levels of 4–6 g dL−1 using apheresis in sheep. Methods: In vitro experiments were performed first to establish the apheresis protocol. In vivo, anesthetized ewes underwent a sham protocol without apheresis (n = 5) or apheresis (n = 4). Both groups were observed for the following six hours at a defined starting point (BE0) to compare Hgb, hematocrit (Hct), coagulation and clinical parameters. For statistical analysis, unpaired t-test with Welch`s correction was used. Results: Hgb levels were effectively decreased by 51 % to mean Hgb of 4.4 g dL−1 in the apheresis group compared to 9.1 g dL−1 in sham (*p < 0.0001). Hct (11.2% vs 25.1 %, *p = 0.01) and RBCs (3.7 vs 8.2 × 106/µl, *p = 0.003) also decreased. The relative number of platelets compared to baseline was different (55.6 ± 10.6% vs. 100 ± 0 %, *p = 0.004), but no hemorrhage was observed. White blood cells (WBCs), lactate, prothrombin ratio and activated partial thromboplastin time (aPTT) remained within similar ranges. Conclusions: Critical normovolemic anemia without bleeding complications was successfully reached by selective RBC depletion in sheep. Investigations of physiological adaptations to severe anemia and pharmaceutical testing can be performed in large animals with depleted RBCs.