The ADVEX advection experiment was one of the most extensive field experiments with the aim to explicitly measure all components of the CO2 mass balance in a control volume, including the advective terms. The main motivation was to find an alternative for correcting nighttime Eddy-covariance (EC) measurements and derived CO2 budgets. This error is usually overcome by removing data measured during periods considered as improper for eddy covariance measurements and, when necessary, replacing them by parameterization or more sophisticated gap-filling. However, this approach has been recently criticized as it suffers from both theoretical and practical shortcomings, and ADVEX was supposed to give a more physically based estimate of Net Ecosystem Exchange (NEE). This relies on the hypothesis that the night flux error mostly results from the emergence of non turbulent transport processes that compete with turbulent flux under low turbulence conditions. Since measuring the non turbulent (horizontal) fluxes requires a three-dimensional approach, and hence additional measurement towers, the complexity and the costs of such an experiment are greatly increased if compared to a single flux tower site. 6 teams from 5 European countries were involved in ADVEX, performing three campaigns in 2005 (Renon/Ritten, Italy) and 2006 (Wetzstein, Germany; Norunda, Sweden) with a duration of 133, 68 and 73 days, respectively. A system of four towers fully equipped with wind velocity, temperature and CO2 concentration profiles and constituting thus a control volume was installed at three European forest sites already equipped with eddy covariance systems as part of the CarboEurope Integrated Project (CEIP) ecosystems flux tower network. These sites are characterised by different topographies. A first presentation of the set up and of the first results was given in Feigenwinter et al. (2008). From CO2 concentrations and wind velocity components determined in the control volume, horizontal and vertical advection were calculated for the control volume. Several advection experiments have been performed during the last decade. However, these field studies are difficult to compare as the results are sensible to the design of the experimental setup and the applied methodology to derive the fluxes from measurements. The ADVEX experiment tried to overcome these limitations by applying an identical experimental setup and methodology for all three measurement campaigns. In this presentation, the most important findings from ADVEX are recapitulated. It is shown, that a night flux correction based on directly measured advective fluxes is hardly practicable (Aubinet et al., 2010). The robustness of advection corrected fluxes was tested against their independence on wind direction and friction velocity. The results show that they vary strongly with these two variables. We also found that advective fluxes of sensible heat were of opposite sign in relation to the advective flux of CO2, but showed a similar evolution in magnitude in relation to these variables. In addition advection measurements were not suitable to be used to determine when conditions for obtaining reliable eddy flux measurements were met. In contrast, larger advection was sometimes observed under strong winds and in flat topography, i.e. in conditions where eddy flux should give reliable estimates of the biotic flux. Nevertheless, site specific case studies showed interesting mechanisms at work in the canopy and sub-canopy spaces. At the alpine slope site Renon/Ritten, advection patterns were closely linked to the local slope wind system and modified by two regional-typical synoptic weather situations. At the Wetzstein site, located on a mountain ridge, the advection patterns for both directions of cross-ridge flows could be clearly related to properties of flows over vegetated hills, namely reverse flow and CO2 accumulation at the slope of the downwind side, leading to positive advection during nighttime. Finally, at the flat site in Norunda, actually foreseen as the “reference site” because of its nearly ideal conditions for EC-measurements, the highest advective fluxes were observed, caused by extremely large horizontal CO2 concentration gradients in the crown space of the forest. Obviously, the ADVEX setup did miss some features of advection related to both, larger (meso) and smaller (turbulence), scales. Many issues related to advection turned out to be extremely complex and we have to conclude that direct measurements of the advective fluxes do not fulfil the expectation for a general solution of the nighttime flux problem. As a main reason, we exhibit the incoherency of the advective fluxes, and of horizontal advection in particular, with the biotic fluxes. It remains to be clarified what the measured fluxes are representative for. The results finally confirm that advection is a scale overlapping process and not restricted to calm and stable nights. At last the question about the sense and a possible design of new advection field experiments arises. Considering the big effort in infrastructure, manpower and time consumption, a cost-benefit analysis would probably do poorly. A new experiment should cover a better resolution in the small scale in the control volume between towers as well as in the mesoscale outside the control volume, which would imply the inclusion of ground based and/or airborne remote sensing techniques like SODAR, RASS, LIDAR and other sensors. If the mass-conservation approach will be applied, additional measurements should be concentrated on the control volume external surfaces.