In Critical Care medicine the concepts of Oxygen Delivery, Oxygen Consumption and Tissue Oxygenation have become fundamental in clinical practice but measurements of Oxygen Content and O2 Transport variables require invasive procedures that could be dangerous for critically ill patients and trigger a septic process. Derived indices obtained combining data from a Blood Gas Analyzer with the data from a multi-wavelength spectrophotometer and using the Ole Siggaard-Andersen pH/Blood Gas computerised algorithm might be the non-invasive answer. On 115 arterial blood samples from critically ill patients, we measured pH, pCO2, pO2, oxygen saturation, total hemoglobin concentration and fractions of carboxy- and methemoglobin. The new algorithm was used to calculate: active hemoglobin concentration, total oxygen concentration, actual half-saturation tension, 2,3-diphosphoglycerate concentration, estimated functional shunt, uncompensated mixed venous pO2 (assuming an arterio-venous oxygen difference of 2.3 mmol/L based on a standard oxygen consumption of 11.2 mmol/min and a standard cardiac output of 4.9 L/min) and the cardiac oxygen compensation factor. In Intensive Care all the oxygen parameters may be determined with sufficient accuracy and precision provided the oxygen saturation level is less than 0.97 and provided the definition of oxygen saturation is properly settled and measurements are performed according to the highest state of the art. However, in critically ill patients in evolution our aim is to maintain an 'optimal' paO2 on the plateau of the Oxygen Dissociation Curve (ODC) and the use of mechanical ventilation, high FIO2, fluid challenges and the rapid improvement of the patient's conditions can cause a value for sO2 greater than or less than 0.97 and an improvement or worsening of the paO2. The p50 calculation both in simultaneously drawn arterial and venous blood permits utilisation of derived indices (pO2uv-, CQ) for sO2 greater than 0.97. The Ole Siggaard-Andersen algorithm seems to give correct p50 values, at high saturation values, particularly when discarding unrealistic values for calculated cDPG. The correlation between p50 calculated by the Ole Sigaard-Andersen algorithm with that derived from classical formula shows the superiority of the findings obtained by means of the new algorithm. In critically ill patients the ODC is usually shifted to the right. The new parameters, pO2uv- and CQ, contain useful informations for clinical practise; but rapid changes in Cardiac Index (CI) and VO2/m2 can be ignored by the new algorithm, if these changes are not associated with a rise in ctO2 or pH and pCO2 changes.(ABSTRACT TRUNCATED AT 400 WORDS)