The commercial Nanoscope I was modified to work with a scanning tunneling microscope (STM) in an open electrochemical cell under potential control. The gold electrode samples were grown epitaxially on green mica. We can simultaneously image the sampe and scan the electrochemical potential. We have studied the Au(111) surface in the following solutions: 0.01 M KCl, 0.01 M, 0.03 M, and 0.05 M AgNO3; 0.01 M NaNO3; 0.01 M and 0.05 M AgClO4. Scanning the potential toward the oxidative range of gold roughens the surface. We found that the roughening was completely reversed when the potential was reduced. In fact, single atom steps on the original surface reappeared unchanged upon completion of an oxidation-reduction cycle. However, this complete reversibility held only so long as a certain electrochemical potential in the solution was not exceeded. In our STM studies we also found a dependence of the roughness amplitude on solution concentration. On samples of the same batch the final roughness was about 40 and 100 Å in 0.01 and 0.05 M AgNO3, respectively. This roughness is much too high to be explained by a monolayer of oxide and therefore it is rather surprising, especially in the case of nitrate and perchlorate electrolytes. The irreversible roughening previously mentioned 0.9 V vs Ag/AgCl in KCl can be explained by dissolution of Au chloride ions, while the reversible part may be understood by creation of neutral adsorbed AuCl. We believe that in general the reversible roughening we observed is probably due to slow adsorption processes that take several minutes to reach equilibrium; without STM images simultaneous with voltammograms, this process would likely have continued to go undetected. © 1991, American Chemical Society. All rights reserved.