An experiment using three New Zealand herds and a total of 632 cows, examined the effect of localised prophylactic treatments with antibiotic at drying-off on the incidence of new intramammary infection during the dry period and at calving. Antibiotic was infused either into the teat canal (0·22 g of dry-cow formulation) or the teat sinus (3·1 g of lactating-cow formulation) of uninfected quarters to eliminate any bacteria present in these locations at the last milking of lactation. These treatments were compared with a negative control (nil treatment) and a positive antibiotic control (infusion of 3·6 g of dry-cow formulation). All antibiotic formulations used the same active ingredient, sodium cloxacillin. No significant reduction in new dry period clinical mastitis was observed for the two localised treatments whereas the positive control treatment achieved 100% reduction in new clinical mastitis compared with untreated control quarters. A 41% reduction (P<0·05) in new Streptococcus uberis infections at calving was associated with the teat canal antibiotic treatment, compared with an 82% reduction (P<0·001) for the positive antibiotic control. Both localised treatments showed a reduced incidence of new intramammary infection (P<0·001) when pooled across periods and pathogens. Teats receiving either the teat canal antibiotic treatment or a full infusion of long acting dry-cow antibiotic had a lower incidence of open teat canals (P<0·05) at 3 weeks after drying-off.