Eighteen cows with a wide range of quarter peak flow rates (0·35–2·22 kg/min) were inoculated with Streptococcus agalactiae and Str. dysgalactiae 4 mm into each streak canal every 3 d for 12 d. Thirty of the 72 quarters developed intramammary infection. Mean peak flow rate and length of streak canal of those quarters that became infected were 1·26 ± 0·08 kg/min (mean ± SEM) and 11·47 ± 0·41 mm respectively, compared with 1·01 ± 0·05 kg/min and 12·05 ± 0·35 mm for those that remained uninfected. Logistic regression analysis showed that the probability of infection increased significantly with the increase in peak flow rate (P = O·O1). The influence of streak canal length on new infection, after allowing for the effect of peak flow rate, was not significant at the 5% level (P = 0·07), suggesting that there may be an inverse relationship between teat duct length and infection, but that it is of secondary importance to peak flow rate. If increased mass of milk distends the teat duct by raising intramammary pressure, then quarter susceptibility to infection is further compromised. These results strongly suggest that the benefits of reduced infection from mastitis control, achieved despite dramatic increases in milk flow rate and milk yield, are significantly underestimated.