SUMMARY All the infected lactating quarters in sixteen herds were given one of four different intramammary cloxacillin treatments. Each quarter was infused three times at 48-hour intervals with either 0·2 g or 0·6 g doses of sodium cloxacillin in either slow or fast release bases. The four preparations eliminated between 60 and 75 per cent of the 321 infections treated, including all the Str . agalactiae and almost all the Str . dysgalactiae infections. Against Staph . aureus infections, although there was no difference in response between the two dose levels, the slow release preparations were significantly ( P Str . uberis infections there was some indication of a dose effect but this was not significant. Linear regressions relating the percentage response to therapy of the staphylococcal infections with various independent characteristics of the treated cows indicated that staphylococcal infections are less likely to respond if they occur in older cows, in cows with several infected quarters, in cows treated in early lactation or when the Whiteside test scores are high. These relationships explain only some of the considerable differences observed between herds in the proportion of staphylococcal infections responding to therapy.