The influenza D virus (IDV) was discovered less than ten years ago. Increased interest in this virus is due to its nature (RNA virus with high mutation rate), its worldwide circulation in livestock species, its probable role in bovine respiratory disease and its zoonotic potential. Until currently, the establishment of positivity cut‐off of the hemagglutination inhibition (HI) assay was not formalized in field conditions for the detection of antibodies directed against IDV in cattle (i.e. the proposed reservoir). In this study, the positivity cut‐off of the HI assays was formally established (titre = 10) using a receiver operating characteristic (ROC) curve. This information was used to estimate the sensitivity (68.04 to 73.20%) and the specificity (94.17 to 96.12%) of two different HI assays (HI1 and HI2, with two different IDV antigens) relatively to virus micro‐neutralization test (VNT) as reference test. Based on the above characteristics, the true prevalence of IDV was then estimated in Morocco using a stochastic approach. Irrespective of the HI assays used, the estimation of the true prevalence was statistically equivalent (between 48.44% and 48.73%). In addition, the Spearman rank correlation between HI titres and VNT titres was statistically good (0.76 and 0.81 for HA1 and HA2, respectively). The positive (0.82 and 0.79 for HA1 and HA2, respectively) and the negative (0.86 and 0.85 for HA1 and HA2, respectively) agreement indices between results of HI assays and VNT were good and similar. This study allowed for a formal establishment of a positivity cut‐off in HI assays for the detection of antibodies directed against IDV. This information is of prime importance to estimate the diagnostic sensitivity and specificity of the test relatively to the VNT (i.e. the reference test). Using these characteristics, the true prevalence of IDV should be determined in a country.