We present one year of data from the first continuous measurements of carbon dioxide mixing ratio in central London. Measurements were made at an 87 m tower site using the new Vaisala CARBOCAP GMP343 instrument, which was found to provide a compact and inexpensive method for mixing ratio monitoring, in an environment where conventional CO2 sensors could not be accommodated. Measurements were compared with a monitoring site outside London, showing that the city's CO2 "dome" was an order of magnitude smaller than measured at lower levels in other cities. During the night time in the summer, the mixing ratio in central London was found to be significantly lower than at the rural site. This was thought to be explained by the proximity of biogenic sources to the rural sensor, differences between urban and rural mixing heights and/or the interception of a vertical mixing ratio gradient at two different measurement heights. Estimation of the CO2 loading of the air entering the city therefore proved problematic during the summer. As a result, we propose that monitoring of CO2 emission rate using this type of measurement may only be possible in the winter when the influence of these factors is minimal. (C) 2008 Elsevier Ltd. All rights reserved.