Optode sensors can provide detailed information on concentrations of dissolved oxygen, which in turn may be used to quantify variations in net primary productivity. Throughout 2005 and 2006 the performance of commercially available oxygen optodes was examined, one in each year. The optode was part of an autonomous measurement system (FerryBox) on a ferry operating between Portsmouth (UK) and Bilbao (Spain). On crossings during which water samples were collected manually, the optode outputs were compared to measurements of dissolved oxygen made by Winkler titrations. The optodes maintained good stability with no evidence of instrumental drift during the course of a year. Over the observed concentration range (230–330 mM m−3) the optode data were approximately 2% low in both years. By fitting the optode data to the Winkler data the median difference between the optode and Winkler measurements is reduced to less than 1 mM m−3 (0.3%) in both years. The most appropriate calibration factor for 2005 was corrected O2 = Optode O2 × 1.018 and for 2006 the corresponding equation is corrected O2 = Optode O2 × 0.884 + 36.8. The standard deviation (95%) of the difference between the individual Winkler measurements was 5 mM m−3 and 3 mM m−3 in 2005 and 2006 respectively. Calculation of the oxygen saturation anomaly is required for calculation of the air sea exchange of oxygen and net biological production. This calculation requires the use of both salinity and temperature data. Salinity is measured to better than 0.1 so the corresponding error in anomaly is less than 0.2 mM m−3. Distortion of the temperature data is present due to warming of the water pumped to the optode. In winter this warming at the optode may be as great as 0.4 °C. The difference in the pumped water temperature can be corrected for by reference to other measurements of sea surface temperature reducing the error to less than 1 mM m−3.