All guinea pigs infected with living Brucella strain 19 developed antibodies which could be detected on the second week and up to the 5th - 6th week according to the serological test used. At the same time skin test was positive as indicated by ery thema and thickning of the site of injection which persisted up to the 8th week. The injection of dead s. 19 induced the production of antibodies of lower titre and shorter duration, and the skin test was negative. The sera of guinea pigs infected with living rough strain 45/20 revealed the presence of antibodies, detectable by all serological tests, with the exception of Rose Bengal test, only when the antigen used for testing was prepared from the same strain used for infection (45/20). However the use of smooth antigen gave negative results in sera of all animals infected with rough strain. On the other hand, the skin test was positive. The dead rough strain 45/20 evoked the production of antibodies in guinea pigs almost similar to the living strain but the skin test was negative. From these results it can be concluded that the rough strain 45/20 is a promising vaccine as it induces antibodies not detectable with smooth antigen routinely used for serological diagno sis of infection. However, it would be recommended to use the dead vaccine if the application of skin test is desired. The relatively low potency of dead vaccine may be improved by the incorportion of a suitable adjuvant. This point needs further studies.