"The higher the volume of blood cultured the higher the yield of blood cultures" has been a well-accepted dictum since J. A. Washington II performed his classic work. This rule has not been questioned in the era of highly automated blood culture machines, nor has it been correlated with clinical variables. Our objective in this study was to complete a prospective analysis of the relationship between blood volume, the yield of blood cultures, and the severity of clinical conditions in adult patients with suspected bloodstream infections (BSI). During a 6-month period, random samples of blood cultures were weighed to determine the volume of injected blood (weight/density). Overall, 298 patients with significant BSI and 303 patients with sepsis and negative blood cultures were studied. The mean volume of blood cultured in patients with BSI (30.03 +/- 14.96 ml [mean +/- standard deviation]) was lower than in patients without BSI (32.98 +/- 15.22 ml [P = 0.017]), and more episodes of bacteremia were detected with <20 ml (58.9%) than with >40 ml (40.2%) of blood cultured (P = 0.022). When patients were stratified according to the severity of their underlying condition, patients with BSI had higher APACHE II scores, and higher APACHE II scores were related to lower sample volumes (P < 0.001). A multivariate analysis showed that in the group of patients with APACHE II scores of >/=18, higher volumes yielded higher rates of bacteremia (odds ratio, 1.04 per ml of blood; 95% confidence interval, 1.001 to 1.08). We conclude that the higher yield of blood cultures inoculated with lower volumes of blood reflects the conditions of the population cultured. Washington's dictum holds true today in the era of automated blood culture machines.