In dairy cows, the milk production (MP) response to anthelmintic treatment (AT) against gastrointestinal nematodes is highly variable. This study aimed to develop a decision tree based on nested criteria to increase the probability of MP response. A randomized controlled trial was conducted using injectable eprinomectin at housing in 123 French grazing dairy herds. Monthly individual milk yields were obtained. Linear mixed models were used to assess the effect of AT on MP. Starting with all herds (Step 0), the herd level criterion associated with the highest MP response to AT was identified (step 1). Then, among farms meeting this first criteria, a second criteria associated with the highest MP response was identified (step 2). This was repeated a third time on herds meeting the first 2 criteria (step 3). Furthermore, at each step, within the selected herds, the cow characteristics associated with the highest MP gain were identified. When considering all the herds (step 0), the effect of AT was small but significant (+0.3 kg/cow/day). The first criterion (step 1) was the % of grazed grass in the diet: no MP gain in low-pasturing herds versus +1 kg/cow/day in moderate/high-pasturing herds. In this subset (n=50), the second criterion (step 2) was the TEC (Time of Effective Contact with GIN infective larvae before the first calving): +0.6 kg/cow/day in high-TEC herds versus +1.4 kg/cow/day in low-TEC herds. In this new subset (n=17), the third criteria (step 3) was the bulk tank milk Ostertagia ODR: +0.2 versus +1.8 kg/cow/day when ODR < 0.9 and ≥ 0.9, respectively. In steps 0, 1 and 2, best responding cows had calved during the grazing season and/or were low-producing cows within their herd. Those herd and cow-level criteria, used sequentially, can be useful to build a decision tree that increases the probability of MP gains.