Mercury contamination in groundwater has been considered as an environmental and public health issue all over the world. Yet, effective in situ remediation techniques have been lacking. A thiol-modified graphene oxide/Fe-Mn composite (SGO/Fe-Mn) was employed as a reactive sorbent of permeable reactive barrier (PRB) for in situ remediation of mercury contaminated groundwater using fixed-bed columns. Mercury existed as HgCl, Hg(OH), and HgClOH, and was mainly removed through surface complexation. The Brunauer-Emmett-Teller sorption isotherm model provided adequate fitting of the sorption isotherm data with a maximum monolayer sorption capacity of 112.03 ± 16.59 mg g. Breakthrough time, the time when 5% of initial Hg concentration is measured in the effluent, increased with the decrease of influent mercury concentration, pore velocity, dissolved oxygen (DO), and dissolved organic matter (DOM). The resultant column sorption capacity was enhanced at higher influent mercury concentration, lower groundwater pore velocity, lower DOM and DO. Moreover, when the SGO/Fe-Mn was thoroughly mixed with quartz sand in the column, the breakthrough time was increased and the resultant sorption capacity was improved compared to the case that SGO/Fe-Mn was packed between two layers of quartz sand. Mathematically, the Adams-Bohart model satisfactorily reproduced the initial behavior of mercury breakthrough curves (<40 pore volumes). Yan model adequately simulated the breakthrough curves. The results reveal the potential of SGO/Fe-Mn as an efficient PRB reactive material for in situ remediation of mercury in contaminated groundwater.