An ultrasensitive metal-labeled amperometric immunoassay (MLAI) method of proteins is reported, on the basis of multiple cycles of gold label/silver staining and galvanic replacement reactions (GRRs), followed by simultaneous chemical dissolution/cathodic preconcentration of silver for in situ microliter-droplet anodic stripping voltammetry (ASV) detection on the immunoelectrode. Briefly, antibody 1 (Ab 1 ), bovine serum albumin (BSA), antigen, and Au nanoparticles functionalized with antibody 2 (Ab 2 −AuNPs) were successively anchored on a Au-plated glassy carbon electrode (GCE) to form a sandwich-type immunoelectrode (Ab 2 −AuNPs/antigen/BSA/Ab 1 /Au plate /GCE). Silver was selectively stained on the catalytic AuNPs surfaces through chemical reduction of silver cations by hydroquinone (gold label/silver staining). A beforehand "potential control" in air and then injection of 5 μL of 25% aqueous HNO 3 on the immunoelectrode surface for dissolution of the stained silver enabled rapid cathodic preconcentration of atomic silver onto the electrode surface as entirely as possible from the lysate, followed by ASV detection of silver. Multiple GRRs between HAuCl 4 and the stained silver were used to amplify the signal, and the stoichiometry of this GRR at the electrode surface was clarified by an electrochemical quartz crystal microbalance. Under optimized conditions, this method was used for ultrasensitive quantitative analysis of human immunoglobulin G (IgG) and human α-fetoprotein (AFP), giving limits of detection (LODs, S/N = 3) of 0.2 fg mL −1 for IgG and of 0.1 fg mL −1 for AFP (equivalent to five molecules in the 6 μL samples employed for analysis of both proteins), ultrahigh sensitivity, excellent selectivity, and little consumption of reagent. The thermodynamic feasibility of such a single-molecule-level amperometric immunoassay (interface-based bioassay) is theoretically proven on the basis of a deduction of the thermodynamic equilibrium of the immunological reactions occurring at the electrode|solution interface.