Context The impact of gastrointestinal nematode (GIN) infections on the milk production of mature dairy cows has been the subject of scientific research since the 1970s. Most of these studies on dairy cows have been conducted in lowland, mainly pasture-based dairy systems with Holstein cows, but there are also important commercial dairy areas in more marginal, upland and alpine regions. The characteristics of dairy farming in these areas -- low-intensity, transhumance, small herd sizes and using regional breeds -- means that caution needs be applied in extrapolating the results of research to these systems. Two studies have been published in which the effects of treatment with the anthelmintic eprinomectin have been evaluated under alpine conditions. In those studies, animals were not systematically sampled for GIN, and so attributing the increase in milk yield to elimination of GIN could only be speculated upon. On the basis of these previous observations, the major aim of this randomised, controlled field trial was to assess the effect of a treatment with eprinomectin on milk yield, somatic cell counts (SCCs) and milk composition of regularly grazed multiparous dairy cows under alpine grazing conditions. Main conclusion This study confirmed that the effect of eprinomectin treatment on the milk yield of dairy cows, which has been reported in lowland dairy herds, can also be observed in cows that are grazed in alpine regions, and supports the hypothesis that this effect is predominantly due to the elimination of GIN. The effects of treatment on SCC were modest but consistent with some other reports in the literature. Approach The effect of a treatment with pour-on eprinomectin on milk yield, milk composition and SCC was studied in 105 dairy cows kept on seven farms in South Tyrol, Italy. The farms were chosen according to their history of pasturing the dairy cattle during the grazing season. On each farm, half of the animals were treated with eprinomectin and the other half formed an untreated control group. Allocation of cattle to one of the two groups was random. Three test-day records per animal were obtained before treatment (on a median days -117, -75 and -33) and another three testday records were obtained after treatment (median days 22, 62 and 131). Test-day records comprised milk yield, milk composition, SCC and days in milk (DIM). On the day of treatment, faecal samples were taken from the rectum for a coprological analysis for GIN, Moniezia species, Eimeria species and Fasciola hepatica, and blood samples were taken at the same time for F hepatica serology. Mixed-effects models to model hierarchical data (farm and cow within farm) were computed to examine the effects of several predictors of milk yield, milk fat and milk protein. The models contained the dichotomous information about when the treatment was applied relative to the test day and dichotomous information about the presence of coprologically detected GIN as fixed effects at the categorical scale, and days in milk (DIM), DIM², age (in months) and 305-day milk yield predicted from the previous lactation as fixed effects at the continuous scale. Results There were no statistically significant differences in the prevalence of GIN, Eimeria species, Moniezia species and F hepatica (in faeces as well as seroprevalence) between cattle allocated to the treatment or control group. Overall, half of the cows had positive GIN faecal egg counts and Eimeria species oocysts were found in over half of the faecal samples. Tapeworm segments and eggs were also observed in some samples. Positive egg counts for liver fluke were seen in 25 per cent of the animals, and 58 per cent were positive serologically. There were highly significant positive effects of treatment on the second test day (day 62:+1.90 kg, P=0.002) and third test day (day 131: +2.63 kg, P< 0.001) after treatment compared with the control animals. SCC was significantly lower (P=0.020) in the treated animals on the second post-treatment test day. Interpretation and notes of caution The increase in milk yield after treatment with eprinomectin might have been due to a reduction in GIN infection, as the presence of gastrointestinal strongyles was estimated to significantly lower milk production by 0.90 kg/day over the entire study period. Given that only half of the animals were treated and that in these animals only half of the testday records were recorded after treatment, the estimates for GIN infection reflect the estimates for the overall effect of the eprinomectin treatment accurately. The prevalence of GIN at the time of treatment was similar in the control and treatment groups. The two groups were not perfectly balanced for age and days in milk, mainly due to the farm structures in the trial area. To minimise potential bias, days in milk and age were included into the multivariate mixedeffects models. Whether the fact that treated animals were younger than control animals influenced the effect of treatment on milk yield and, if so, whether this resulted in an under- or overestimation cannot be determined. On the one hand, one could argue that older cows have generally been more exposed to GIN than younger cows during their lifetime, resulting in better immunity to GIN in older cows, which might lead to an overestimation of the treatment effect if treated cows are younger. On the other hand, younger animals yield less than older animals, which would result in an underestimation of the treatment effect if one would assume that the treatment effect is proportional to the total milk yield. Significance of findings This study revealed a positive effect of eprinomectin treatment at autumn housing on the milk yield of dairy cattle that grazed on alpine pastures during the summer months (May to October). Because the presence of gastrointestinal strongyles was estimated to lower milk production significantly over the entire study period, this study supports the hypothesis that the effect was predominantly due to the elimination of GIN. [ABSTRACT FROM AUTHOR]