A Scottish isolate of Piscirickettsia salmonis (SCO-95A), previously shown by intraperitoneal injection to have a lethal dose (LD50) of < 2 x 10(3) infectious rickettsial units, was tested for virulence by bath challenge, surface application to the skin, or dorsal median sinus injection. Atlantic salmon Salmo salar post-smolts were used in all experiments, and exposure to 1 x 10(5) tissue culture infective doses (TCID) of P. salmonis ml(-1) for 1 h in a bath challenge resulted in only 1 mortality, 18 d later, in 10 exposed fish. Application of 2.5 x 10(6) TCID of P. salmonis SCO-95A to paper discs on the skin failed to induce any mortalities within 42 d. Intraperitoneally, fish were administered vaccines containing 10(9) heat-inactivated (100 degrees C, 30 min) or 10(9) formalin-inactivated P. salmonis SCO-95A in adjuvant, with a control group receiving phosphate-buffered saline (PBS) in adjuvant. After an induction period of over 6 mo fish were challenged by injection of P. salmonis into the dorsal median sinus. Mortalities in the control group reached 81.8% and the heat-inactivated and formalin-inactivated vaccines gave significant protection from P. salmonis, with relative percentage survivals of 70.7 and 49.6%, respectively. The nature of the protective antigen is unknown, but could be lipopolysaccharide or a heat-stable outer membrane protein. Fish that survived a dorsal median sinus challenge of P. salmonis or were cohabitants showed a strong immune response to P. salmonis.