Haptoglobin (Hp) is an emerging diagnostic marker in cattle, and knowledge of suitable sample types and measurement methods is important. The aims of this study were to compare the results of a colorimetric assay (CA) and an ELISA for bovine Hp using serum, EDTA plasma, and lithium-heparinized (LH) plasma, respectively, and to assess the diagnostic potential for puerperal metritis. In experiment 1, Hp was measured in pooled aliquots of serum (n = 10), EDTA plasma (n = 10), and LH plasma (n = 10) of 100 healthy fresh lactating dairy cows from 10 farms using both the CA and the ELISA. In experiment 2, five healthy and five cows with acute puerperal metritis were sampled, and Hp was determined using both assays for all three sample types. In experiment 3, aliquots of serum and LH plasma from cows in different lactation stages were transferred into plain, EDTA-coated, and LH-coated tubes and mixed before colorimetric analyses. Distilled water was also placed into each tube type and treated similarly. Plasma samples measured with the CA showed on average 2.3 (EDTA) and 2.5 (LH) times higher Hp concentrations compared with serum, whereas no differences were seen with the ELISA results between sample types. Based on a clinical cut-off value, both methods differentiated sick from healthy cows. Haptoglobin measurements with the ELISA were less precise compared with CA measurements due to high dilutions. No influence of the anticoagulants on the CA was observed. Due to measurement discrepancies between serum and plasma, CAs for bovine Hp based on peroxidase activity should be performed with serum, or specific reference ranges for plasma samples should be established. In this study, CA results obtained with LH plasma were more precise than results obtained with EDTA plasma. Both the CA and the ELISA are suitable diagnostic tools for the diagnosis of puerperal metritis, but CA measurements were more precise in this study.