Three groups of dairy cows were immunized by subcutaneous (s.c.) administration of a preparation of live Streptococcus uberis (strain 0140J) and an intramammary infusion of a soluble surface extract derived from same the bacteria. Animals in Groups 1 and 2 received two s.c. vaccinations plus an intramammary inoculation. Animals in Group 3 received two s.c. vaccinations but did not receive the intramammary infusion. In addition to the vaccinated animals, each group also contained two non-vaccinated (control) animals. All animals were challenged experimentally by intramammary infusion (in two quarters per animal) of ca 100 c.f.u. of S. uberis (strain 0140J or C221) and monitored for clinical signs of disease, bacterial numbers in milk, somatic cell count in milk, and daily milk yield for the following 10 days. Animals in Group I were challenged with strain 0140J. Only one out of six challenged quarters of three vaccinated cows developed clinical disease compared to all (four out of four) quarters of non-vaccinated cows. Animals in Group 2 were challenged with strain C221. All challenged quarters of three vaccinated (six out of six) and two non-vaccinated (four out of four) cows developed clinical mastitis. Animals in Group 3 were challenged with strain 0140J. Five out of eight quarters on four vaccinated cows developed clinical mastitis but the onset was delayed in comparison with that in both non-vaccinated cows in which four out of four challenged quarters developed clinical mastitis. These results indicated that vaccination with live S. uberis protects against challenge with the homologous strain but was less effective against a heterologous strain. Reduced protection was also seen when the intramammary booster was omitted.