Empirical, geostatistical and geochemical modeling techniques were used to determine whether chemical, isotopic and mineralogical data collected in surface-based borehole investigations at the Horonobe Underground Research Laboratory (URL) site in Hokkaido, Japan were representative of the natural environment before it was disturbed by borehole drilling, hydraulic testing and sampling. Water samples collected either by pumping groundwaters to the surface from borehole sections that had been isolated by inflatable packers or by squeezing porewaters from drillcores sampled from the same borehole and sampling depth were found to be similar in certain respects (m(Na+) m(Cl-), delta(18)O and delta D) but not others (m(SO42-), m(Ca2+), m(Mg2+) and m(K+)). The inconsistencies suggest that a majority of the porewater samples were contaminated as a result of pyrite oxidation and dissolution of carbonate minerals during sampling, storage and/or squeezing of the drillcores. Multivariate geostatistical analyses support this hypothesis, and further suggest that uncontaminated porewaters and groundwaters can be divided into three representative groups: a low salinity, low delta(18)O and low delta D group; a moderate salinity, low Ca-Mg group: and a high salinity, high K and moderate Ca-Mg group. The groundwaters also contain high concentrations of dissolved gases, including CH(4)(g), CO(2)(g), H(2)S(g) and various hydrocarbons, which exsolve from the groundwaters as they are pumped to the surface for sampling. The effects of such degassing on the chemistry of groundwater samples was evaluated by using a reaction-path model to simulate the titration of gases collected at the surface back into the samples. Results suggest that undisturbed groundwaters are slightly more acidic than their sampled counterparts, and contain roughly equivalent activities of dissolved CH(4)(aq) and carbonate species. Redox potentials calculated using the corrected groundwater compositions and assuming equilibrium for the CH(4)(aq)/CO(2)(aq) redox couple are significantly more negative than those determined in situ in one borehole using a downhole chemical probe, and those that can be inferred from the ubiquitous presence of small amounts of framboidal pyrite and siderite in siliceous biogenic sediments of the Horonobe area. The empirical, geostatistical and geochemical modeling techniques considered in this study can be adapted for use in characterizing the hydrogeochemical environment of a site that will eventually be selected to host a geologic repository for high-level nuclear waste in Japan. (C) 2011 Elsevier Ltd. All rights reserved.