Background Clinical reports on brain tissue oxygen tension differ in suggested threshold values for defining cerebral ischemia using the Licox and Neurotrend/Paratrend system. We evaluated in vitro performance of both first and second generation devices. Materials and methods Response rate and accuracy in solutions with oxygen tensions from 0 to 150 mm Hg were measured. Findings Ninety-five percent Response times were 102 +/- 13 seconds for first generation Licox probes and 13 5 +/- 24 s for Paratrend (n=6, each probe), with second generation probes at 134 +/- 4 and 116 +/- 16 s respectively. At pO(2) 150 mmHg Licox and Paratrend probes were accurate with 2.2% and 2.1% error, respectively and 2.6% and 4.1% for later generation. At pO(2) 18 mmHg, Paratrend overestimated by 16.5% (absolute error range 2.18 to 4.18 mmHg), 7.4% for Neurotrend, Licox underestimated by 1.8% (absolute error range 0.08 to 0.52 mmHg) with 3.6% for the second generation probe. Conclusions Differences between the first generation probe types, while statistically significant (p<0.001), may not be clinically relevant. Overestimation Of pO(2) by Neurotrend and small underestimation by Licox partially explain differences in published thresholds for cerebral ischemia. The Neurotrend was slightly more accurate and faster than the Paratrend system.