In the early days, acid-base studies were performed on venous blood. Venous specimens are nearly as accurate as arterial blood for pH and HCO3 (or PCO2) measurements if blood is obtained anaerobically from a motionless hand or arm before the tourniquet is released. Nevertheless, arterial specimens have mostly replaced venous ones because venous blood provides less accurate data in some conditions such as decreased tissue perfusion due to shock. Even more important, one can also obtain blood oxygen measurements (PO2) with arterial samples. Arterial blood is most often drawn from the radial artery in the wrist. Arterial puncture is little more difficult than venipuncture, and there is a small but definite risk of extravasation and hematoma formation that could compress the artery. Although glass syringes have some technical advantages over plastic syringes or tubes (such as a slightly smaller chance of specimen contamination with room air than when using plastic syringes), most hospitals use only plastic. It is officially recommended that the specimen tube or syringe should be placed in ice immediately for transport to the laboratory, both to prevent artifact from blood cell metabolism and to diminish gas exchange between the syringe and room air. The blood must be rewarmed before analysis. Actually, it is not absolutely necessary to ice the specimen in most cases if the analysis takes place less than 15 minutes after the specimen is obtained. Icing the specimen in plastic tubes can elevate PO2 values a little if they are already over 80 mm Hg (10.7 kPa). One investigator found that at 100 mm Hg (13.3 kPa), the false elevation averages 8 mm Hg (1.06 kPa). Also, icing in plastic tubes increases plasma viscosity over time and interferes with resuspension of the RBC, which affects hemoglobin assay in those instruments that calculate O2 content from PO2 and total hemoglobin. If mixing before assay is not thorough, hemoglobin values will be falsely decreased somewhat. In addition, if electrolytes are assayed on the arterial specimen, potassium may be falsely increased somewhat.

Capillary blood specimens from heelstick are often used in newborn or neonatal patients because of their small blood vessels. Warming the heel produces a semiarterial (“arterialized”) specimen. However, PO2 is not reliable and PCO2 sometimes differs from umbilical artery specimens. The majority of reports do not indicate substantial differences in pH; however, one investigator found a mean decrease in PCO2 of 1.3 mm Hg (0.17 kPa), a mean pH increase of 0.02 units, and a mean decrease of 24.2 mm Hg (3.2 kPa) in PO2 from heelstick blood compared to simultaneously drawn umbilical artery blood.

Heparin is the preferred anticoagulant for blood gas specimens. The usual method is to wash the syringe with a heparin solution and then expel the heparin (which leaves about 0.2 ml of heparin in the dead space of the syringe and needle). If too much heparin remains or the blood sample size is too small (usually when the sample is <3 ml), there is a disproportionate amount of heparin for the amount of blood. This frequently causes a significant decrease in PCO2 (and bicarbonate) and hemoglobin values, with a much smaller (often negligible) decrease in pH. These artifactual decreases in PCO2 are especially apt to occur when the sample is obtained from indwelling catheters flushed with heparin.