BACKGROUND: The use of hypertonic crystalloid solutions, including sodium chloride and bicarbonate, for treating severe sepsis has been much debated in previous investigations. We have investigated the effects of three crystalloid solutions on fluid resuscitation in severe sepsis patients with hypotension. METHODS: Ninety-four severe sepsis patients with hypotension were randomly assigned to three groups. The patients received the following injections within 15 min at initial treatment: Ns group (n = 32), 5 ml/kg normal saline; Hs group (n = 30), with 5 ml/kg 3.5% sodium chloride; and Sb group (n = 32), 5 ml/kg 5% sodium bicarbonate. Cardiac output (CO), systolic blood pressure, mean arterial pressure (MAP), body temperature, heart rate, respiratory rate and blood gases were measured. RESULTS: There were no differences among the three groups in CO, MAP, heart rate or respiratory rate during the 120 min trial or the 8 hour follow-up, and no significant differences in observed mortality rate after 28 days. However, improvement of MAP and CO started earlier in the Sb group than in the Ns and Hs groups. Sodium bicarbonate increased the base excess but did not alter blood pH, lactic acid or [HCO3]- values; and neither 3.5% hypertonic saline nor 5% sodium bicarbonate altered the Na+, K+, Ca2+ or Cl- levels. CONCLUSION: All three crystalloid solutions may be used for initial volume loading in severe sepsis, and sodium bicarbonate confers a limited benefit on humans with severe sepsis. TRIAL REGISTRATION: ISRCTN36748319.