The kinetics of adsorption and competitive desorption of wild-type streptavidin (WT SA) and three genetically engineered mutants (S27A, N23E, and W120A) was studied at gold surfaces functionalized with mixed alkylthiolates, some terminated with biotin headgroups and the rest with oligo(ethylene oxide) using surface plasmon resonance (SPR). The saturation coverage of the protein varied strongly with surface biotin concentration (X-BAT) and was independent of mutation (except at very low and very high X-BAT, where a weak dependence was seen). Initial adsorption rates were nearly diffusion-limited except at extremely low X-BAT, where the rate varied weakly between mutants in accordance with their differing strengths of binding to biotin. Initial sticking probabilities were estimated to be between similar to1-6 x 10(-6) per collision with the surface. The adsorbed SA desorbs upon introduction of solution-phase biotin. For X-BAT below 1%, the desorption rate constants of the SA variants closely follow their off-rate constants measured in homogeneous solution (which at 25 degreesC are WT = 4 x 10(-6) sec(-1), N23E = 1.6 x 10(-3) sec(-1), S27A = 1.2 x 10(-3) sec(-1) and W120A estimated to be ca. 23 s(-1)). This proves that SA is mainly bound to the surface by a single Sk biotin link at very low XBAT Importantly, for X-BAT between 10 and 40%, where desorption is 30- to >1000-fold slower and the saturation coverage maximizes, the ratios of off-rate constants between mutants (W120A/N23E and W120A/S27A) are approximately the square of their ratios for XBAT below 1%. This squaring strongly suggests that the dominant species at these coverages is doubly bounded SA (i.e., immobilized via two surface biotins). The kinetics are explained with a mechanism involving only two first-order rate constants, that is, for (1) the slow dissociation of any bond between an SA site and a surface-immobilized biotin and (2) the fast reforming of this bond in the special case that it was released from a doubly bonded SA whose other site is still linked to one surface-immobilized biotin. The rate constant for (2) is almost independent of the SA mutant, as it is for adsorption. For X-BAT > 60%, the desorption rates again approach the singly bound SA values, and the ratios of rate constants for the SA variants drop to slightly less than below 1% biotin. This is due to the dominance of singly bonded SA, plus a contribution from nonspecific binding, consistent with structural studies of these alkylthiolate films.