The net peak-current response as a function of frequency, for both direct and stripping square-wave voltammetry at mercury film electrodes, can differ significantly from theory when either the surface morphology of the mercury film deviates from being a true flat film or the substrate interacts with the analyte in some manner. Mercury layers of various morphologies (film, patches or droplets) and thicknesses (1.0 to 24.0 μm) were prepared by depositing Hg(II) onto an iridium or glassy carbon substrate. Test solutions of Cd(II), Pb(II) and K 4Fe(CN) 6 in 0.1 M potassium nitrate were used. For square-wave frequencies of 10 to 2000 Hz, the iridium electrodes with true mercury films and the bare Ir electrodes gave a theoretical response for both direct SWV and SWASV with all analytes tested. For the Hg films on glassy carbon or the non-solid Hg-film on Ir, the peak-current response for direct SWV began to decrease at SW frequencies of 900 and 100 Hz respectively. The effect with SWASV was more pronounced, the decrease occurring at 100 and 50 Hz. This behavior is attributed to several factors but mainly to the increased exposure of the substrate as a result of the decreasing thickness of the diffusion layer, and the behavior of the mercury droplets as a collection of ultramicroelectrodes.