Vibrio vulnificus is Gram-negative bacterium that contaminates oysters, causing highly lethal sepsis after consumption of raw oysters and wound infection. We previously described two sets of V. vulnificus strains with different levels of virulence in subcutaneously inoculated iron dextran-treated mice. Both virulent, clinical strains and attenuated, environmental strains could be recovered in high numbers from skin lesions and livers; however, the attenuated environmental strains required significantly higher numbers of colony-forming units (cfu) in the inoculum to produce lethal infection. Using some of these strains and an additional clinical strain, we presently asked if the different abilities to cause infection between the clinical and environmental strains were due to differences in rates of growth or death of the bacteria in the mouse host. We therefore constructed a marker plasmid, pGTR902, that functions as a replicon only in the presence of arabinose, which is not present in significant levels in animal tissues. V. vulnificus strains containing pGTR902 were inoculated into iron dextran-treated and untreated mice. Measuring the proportion of bacteria that had maintained the marker plasmid recovered from mice enabled us to monitor the number of in vivo divisions, hence growth rate; whereas measuring the number of marker plasmid-containing bacteria recovered enabled the measurement of death of the vibrios in the mice. The numbers of bacterial divisions in vivo for all of the strains over a 12-15 h infection period were not significantly different in iron dextran-treated mice; however, the rate of death of one environmental strain was significantly higher compared with the clinical strains. Infection of non-iron dextran-treated mice with clinical strains demonstrated that the greatest effect of iron dextran-treatment was increased growth rate, while one clinical strain also experienced increased death in untreated mice. V. vulnificus inoculated into iron dextran-treated mice replicated extremely rapidly over the first 4 h of infection with doubling times of approximately 15-28 min. In contrast, one of the environmental strains exhibited a reduced early growth rate. These results demonstrate that differences in virulence among naturally occurring V. vulnificus can be explained by diverse abilities to replicate rapidly in or resist defences of the host. The marker plasmid pGTR902 should be useful for examining virulence of bacteria in terms of differentiating growth verses death in animal hosts for most Gram-negative bacteria.