The introduction of biosensor technology for near bedside measurement of plasma lactate concentrations has been a promising step for critical care profiling. However, methodological drawbacks and relevant inaccuracy have been reported. With the advent of a new biosensor (Chiron Diagnostics) and a revised NOVA Biomedical device, accuracy was expected to be improved. The goal of the present investigation was to evaluate the accuracy of both methods. METHODS Two devices (System 860, Chiron Diagnostics; StatProfile 9, NOVA Biomedical) were simultaneously analysed using 9 biosensors in both fresh frozen plasma and citrated whole blood. The results were compared with an established photometric method (Lactat PAP, Analyticon). Measurements were performed as duplicates (n = 1120) before and after the addition of 1 molar sodium lactate solution (2-24 mmol/L). For the estimation of between-day precision commercially available aqueous and serum-based quality controls were analysed daily over a period of 60 days. RESULTS Reproducibility in blood was 2.6 +/- 2.8% (Chiron), 4.1 +/- 4.0% (NOVA) and 1.5 +/- 2.1% (Analyticon), in plasma respectively 2.1 +/- 2.4%, 2.1 +/- 2.9% and 1.0 +/- 1.1%. Mean inaccuracy in plasma presented to be -0.2 +/- 16.4% (plasma) and +7.2 +/- 13.1% (blood) for Chiron, +9.4 +/- 18.4% and +18.7 +/- 16.7% for NOVA, and -37.8 +/- 18.2% and -27.5 +/- 17.6% for Analyticon. Calculated between-day-precision (variation coefficients mean values) was 11.5 +/- 4.9% (Chiron) and 14.0 +/- 5.9% (NOVA). CONCLUSIONS Although accuracy of lactate concentrations obtained with biosensor technology has improved (mean 0-18%), the variability of the results still poses a problem (mean 13-18%). Therefore, from the methodological point of view, interpretation of a single lactate value requires caution when applying to the critically ill, particularly with view to threshold values, and should be considered vis-à-vis other options.