Euhedral pyrite crystals in 46 open bedrock (granitoid) fractures at depths down to nearly 1 km were analysed for sulphur isotope ratios (delta S-34) by the in situ secondary ion mass spectrometry (SIMS) technique and by conventional bulk-grain analysis, and were compared with groundwater data. Twenty nine of the fractures sampled for pyrite had corresponding data for groundwater, including chemistry and isotopic ratios of sulphate, which provided a unique opportunity to compare the sulphur-isotopic ratios of pyrite and dissolved sulphate both at site and fracture-specific scales. Assessment of pyrite age and formation conditions were based on the geological evolution of the area (Laxemar, SE Sweden), and on data on co-genetic calcite as follows: (1) the isotopic ratios of the calcite crystals (delta O-18, delta C-13, Sr-87/Sr-86) were compared with previously defined isotopic features of fracture mineral assemblages precipitated during various geological periods, and (2) the delta O-18 of the calcites were compared with the delta O-18 of groundwater in fractures corresponding to those where the calcite/pyrite assemblages were sampled. Taken together, the data show that all the sampled fractures carried pyrite/calcite that are low-temperature and precipitated from the current groundwater or similar pre-existing groundwater, except at depths of -300 to -600 m where water with a glacial component dominates and the crystals are from pre-modern fluids. An age of <10 Ma are anticipated for the pre-modern fluids. The delta S-34(pyr) showed huge variations across individual crystals (such as -32 to +73 parts per thousand) and extreme minimum (-50 parts per thousand) and maximum (+91 parts per thousand) values. For this kind of extreme S-isotopic variation at earth-surface conditions there is no other explanation than activity of sulphur reducing bacteria coupled with sulphate-limited conditions. Indeed, the most common subgrain feature was an increase in delta S-34(pyr) values from interior to rim of the crystal, which we interpret are related to successively higher delta S-34 values of the dissolved source SO42- caused by ongoing bacterial sulphate reduction in fractures with low-flow or stagnant waters. The measured groundwater had delta S-34(SO4) values of +9 parts per thousand to +37 parts per thousand, with the highest values associated with low sulphate concentrations. These values are overall, and especially in the sulphate-poor waters down to -400 m, somewhat higher than the anticipated initial values, and can thus, like for the S-34-enriched pyrites, be explained by a Rayleigh distillation process driven by microbial sulphate reduction. An intriguing feature was that the delta S-34(SO4) values of the groundwater were in no case reaching up to the values required to produce biogenic pyrite with delta S-34 values of +40 parts per thousand to +91 parts per thousand. To explain this feature, we suggest that groundwater in low-flow fractures with near-stagnant water (carrying sulphate and pyrite with high delta S-34) is masked by high-flow parts of the fracture system carrying groundwater that often contains sulphate in abundance and considerably less fractionated with respect to S-34 and S-32. In order to gain detailed knowledge of chemical processes and patterns in groundwater in fractured rock, fracture-mineral investigations are a powerful tool, as we have shown here for the sulphur system. (C) 2012 Elsevier Ltd. All rights reserved.