We thank Drs. Cole and Singh and colleagues for their thoughtful comments. We agree on most points. Decisions regarding fluid management are among the most difficult in clinical medicine in general and in the ICU in particular. We believe that this decision should be based on an indepth understanding of the disease process and treatment strategy, with a review of fluid balance, oxygenation index, urine output, chest radiograph, and renal function (and electrolytes), together with a marker of volume responsiveness. We believe that dynamic parameters such as stroke volume variation are currently the best markers of volume responsiveness.1 While the models of Guyton and Starling are magnificent in their simplicity, therapeutic interventions based on these principles may lead to therapeutic errors in complex critically ill patients. This is demonstrated by the fact that no clinical study (of which we are aware) has shown central venous pressure to be a reliable tool in the fluid management of critically ill or injured patients. Indeed, the data suggest that the reliability of central venous pressure for predicting fluid responsiveness is no better than flipping a coin. While the study by Rivers et al2 (and the Surviving Sepsis Campaign)3 are often quoted to support the use of central venous pressure as the preferred goal for fluid resuscitation (at least in patients with sepsis), it should be pointed out the same central venous pressure goals were utilized in both the intervention and control groups.2 Should the data by Rivers et al2 be valid, it should be noted that the control group has the highest reported mortality of any sepsis study.4,5 Based on the current data, we believe that guiding therapy based on central venous pressure is misguided, will lead to serious errors and should be abandoned (at least until supporting data are published).