We evaluated the performance of the Advia 2120 (Siemens) differential leukocyte count (A-Diff) compared to the manual method (M-Diff) in rabbits. EDTA-anticoagulated blood samples collected for diagnostic purposes were analyzed within 6 h of collection. The M-Diff was performed blindly by 2 observers on blood smears by counting 200 cells. We initially included 117 samples; 25 samples were excluded because of suboptimal gating of leukocytes in the Advia peroxidase cytogram or poor blood smear quality. The correlation between the A-Diff and M-Diff was very high for heterophils (r = 0.924,  < 0.001) and lymphocytes (r = 0.903,  < 0.001), high for basophils (r = 0.823,  < 0.001), moderate for monocytes (r = 0.645,  < 0.001), and low for eosinophils (r = 0.336,  = 0.001). The Passing-Bablok regression analyses revealed a small-to-moderate constant error for lymphocytes and a slight constant error for basophils. Small proportional errors were detected for heterophils, lymphocytes, and eosinophils. The Bland-Altman analyses revealed that the Advia significantly underestimates heterophils and overestimates lymphocytes compared to M-Diff. The biases for the other leukocytes were minimal and likely clinical insignificant; however, our results, particularly for eosinophils, should be interpreted cautiously given the observed low percentages in our samples. Given the observed biases in heterophil and lymphocyte percentages in the Advia 2120 CBC results in rabbits, method-specific reference intervals should be used. The Advia can recognize leporine basophils. Evaluation of blood smears is still recommended to investigate abnormal results and erroneous cytograms reported by the Advia.