<p>Carbon dioxide (CO<sub>2</sub>) emissions from running waters represent a key component of the global carbon cycle. However, quantifying CO<sub>2</sub> fluxes across air-water boundaries remains challenging due to practical difficulties in the estimation of reach-scale standardized gas exchange velocities (k<sub>600</sub>) and water equilibrium concentrations. Whereas craft-made floating chambers supplied by internal CO<sub>2</sub> sensors represent a promising technique to estimate CO<sub>2</sub> fluxes from rivers, the existing literature lacks of &#160;rigorous &#160;comparisons &#160;among &#160;differently &#160;designed chambers and deployment techniques. Moreover, as of now the uncertainty of k<sub>600</sub> estimates from chamber data has not been evaluated. &#160;Here, these issues were addressed analyzing the results of a flume experiment carried out in the Summer of 2019 in the Lunzer:::Rinnen - Experimental Facility (Austria). During the experiment, 100 runs were performed &#160;using two different chamber designs (namely, a Standard Chamber and a Flexible Foil chamber with an external floating system and a flexible sealing) and two different deployment modes (drifting and anchored). The runs were performed using various combinations of discharge and channel slope, leading to variable turbulent kinetic energy dissipation rates (1.5 10<sup>-3</sup>< &#949; < 1 10<sup>-1</sup> m<sup>2</sup> s<sup>-3</sup>). Estimates of gas exchange velocities were in line with the existing literature (4 < k<sub>600</sub> < 32 m d<sup>-1</sup>), with a general increase of k<sub>600</sub> for larger turbulent kinetic energy dissipation rates. The Flexible Foil chamber gave consistent k<sub>600</sub> patterns in response to changes in the slope and/or the flow rate. Moreover, Acoustic Doppler Velocimeter measurements indicated a limited increase of the turbulence induced by the Flexible Foil chamber on the flow field (22 % increase in &#949;, leading to a theoretical 5 % increase in k<sub>600</sub>).<br>The &#160;uncertainty &#160;in &#160;the &#160;estimate &#160;of &#160;gas &#160;exchange &#160;velocities &#160;was &#160;then estimated &#160;using &#160;a &#160;Generalized Likelihood Uncertainty Estimation (GLUE) procedure. Overall, uncertainty in k<sub>600</sub> was moderate to high, with enhanced uncertainty in high-energy setups. For the anchored mode, the standard deviations of k<sub>600</sub> were between 1.6 and 8.2 m d<sup>-1</sup>, whereas significantly higher values were obtained in drifting mode. Interestingly, for the Standard Chamber the uncertainty was larger (+ 20 %) as compared to the Flexible Foil chamber. &#160;Our study suggests that a Flexible Foil design and the anchored deployment might be useful techniques to enhance the robustness and the accuracy of CO<sub>2</sub> measurements in low-order streams. Furthermore, the study demonstrates the value of analytical and numerical tools in the identification of accurate estimations for gas exchange velocities.<br>These findings have important implications for improving estimates of greenhouse gas emissions and reaeration rates in running waters.</p>