Of 4,289 episodes of bacteremia detected in 3,631 patients, septic shock was diagnosed in 453 episodes (10.5%). In 56% of shock episodes, septic shock developed more than 24 h after the first positive blood culture was taken. In a logistic regression analysis, variables predictive of septic shock were: advanced age [odds ratio (OR) of 1.015 for an increment of 1 year]; renal failure as an underlying disorder (OR = 1.47); neutropenia (OR of 2.26); curtailed functional capacity (OR of 1.54 for an increment of 1 category); unknown source of infection (OR = 1.66); anaerobic (OR = 2.86), polymicrobial bacteremia (OR = 1.54), or pathogens other than Streptococcus viridans (OR = 0.08 for Streptococcus viridans). The in-hospital mortality associated with septic shock was 80% vs 21% in episodes of bacteremia without shock, and shock episodes accounted for 31% of all deaths. The fatality rate in shock patients given appropriate empiric antibiotic treatment was 74.9% vs 84.7% in patients given inappropriate treatment (p = 0.01). Judging by the present results, host factors are more important determinants for development of septic shock in bacteremic patients than the type of pathogen. Even in patients with shock, appropriate empiric antibiotic treatment was associated with an improved chance of survival.