Herd-level diagnosis of paratuberculosis using a pool-milk ELISA (pool size: n ≤ 50) is a novel, economical, and convenient method to identify blood serological Mycobacterium avium ssp. paratuberculosis (MAP) antibody-positive herds. To date, the diagnostic performance of the pool-milk ELISA has been described only under laboratory conditions where herd prevalence was simulated by the preparation of milk pools consisting of milk samples of cows with a known MAP status determined by fecal culture. In our observational study, test performance under field conditions was studied using pooled milk and individual blood samples. A total of 486 herds within the MAP prevalence reduction program of Lower Saxony, from which pooled milk and individual blood ELISA results were available, were assigned to this study. Data were analyzed for the period between January 1 and December 31, 2018, the first year after herd testing became obligatory in this federal state of Germany. To evaluate whether pooled milk samples reliably distinguish between herds with a MAP-apparent blood serological within-herd prevalence (MAP-Ab-WHPapp) ≥5% and herds with a MAP-Ab-WHPapp <5%, the distribution of the MAP-Ab-WHPapp was compared between pool-positive and pool-negative herds. The MAP-Ab-WHPapp was 3.4% (median; 95% confidence interval = 0-11.4%) in pool-positive herds and 1.2% (median; 95% confidence interval = 0-6.4%) in pool-negative herds. Only 10.8% (n = 12) of the pool sample-negative herds had a MAP-Ab-WHPapp ≥5% and were therefore false negatives, given the aims of the MAP prevalence reduction program. Hence, the pool-milk sampling strategy seems well suited to distinguish between herds with a MAP-Ab-WHPapp ≥ 5% and herds with a MAP-Ab-WHPapp <5% since only 10% of serum MAP-ELISA positive herds were missed. Employing a logistic regression model, we estimated that the minimum blood serological MAP-Ab-WHPapp to detect a pool-positive herd with a probability of 95% was 8%, which fits well with the aim of the MAP prevalence reduction program to focus on herds with a MAP-Ab-WHPapp of ≥5%. Despite the limitations of the control approach, which include milk pool sample collection and a low sensitivity of the ELISA used in milk pools and serum samples, the aims of the MAP prevalence reduction program can be achieved. The results of these field data support that pool-milk sample ELISA is a useful, economical, and low labor-intensive tool to identify herds seropositive for MAP in a MAP prevalence reduction program.