An extra large aperture scintillometer (XLAS) was used over several months across the Thau Lagoon (South of France) to retrieve one-wavelength scintillation and, thence, sensible heat flux. We present the experiment with the XLAS, an eddy-covariance station and meteorological stations measuring on or near the Thau Lagoon. Changes implemented to adapt the scintillometry processing schemes to the above water conditions are presented together with a full error budget, including sensitivity tests to the relevant parameters of the scintillometer processing scheme. The XLAS error budget amounts to 16% (systematic part) +/- 50% (random part). Sensible heat fluxes obtained using the XLAS under unstable atmospheric conditions are then compared to eddy-covariance estimates used as a reference. The scintillometry technique proved to perform satisfactorily in such a watery environment. Some discrepancies observed between the XLAS and eddy-covariance measurements were investigated according to the lagoon fraction of the source area, to discriminate whether they were related to deviations from the Monin-Obukhov similarity theory or to different atmospheric conditions at the respective instrument locations. Local atmospheric conditions agreed well with the Monin-Obukhov similarity theory, especially measurements with source areas largely composed of the lagoon surface. Retaining only the measurements with almost only the lagoon surface in the source area improved the agreement between the XLAS and eddy-covariance measurements. The remaining discrepancies are interpreted as being due to significant location differences between the two instruments, resulting in different atmospheric conditions, and to size differences in the source areas.