Carbon capture and storage in the deep subsurface is proposed as a mitigating solution for reducing anthropogenic CO2 emissions in the atmosphere. The efficiency for mineral storage in mafic rocks is being evaluated by the CARBFIX consortium (https://www.or.is/en/projects/carbfix) at the Hellisheidi pilot site (Iceland). In 2012, two injections were performed (i.e. pure CO2 or H2S/CO2/H2 mix from gas separation station associated to the power plant). The last injection was interrupted because of a decrease of one order of magnitude of the injectivity index. Back-flushing of the injection well by airlift pumping was then performed for remediation. We have studied the microbial diversity and the mineralogy of the collected materials in order to understand the causes of clogging. Results revealed an increase of gene copy numbers (detected by qPCR) and a loss of diversity, in comparison to pre-clogging communities. The endemic Betaproteobacteria sulfur-oxidizing Thiobacillus sp. and the Deltaproteobacteria sulfur-reducing Desulfurivibrio sp. were the main representatives of the bacterial communities, as revealed by 454-pyrosequencing. Microscopic observations using Scanning Electron Microscopy and Raman spectroscopy evidenced that the injected gaseous H2S reacted with the Fe-bearing basalt minerals to abiotically form submicrometric Fe-sulfides, then oxidized by Thiobacillus that formed to adhere to this substrate, a dense biofilm aggregating individual Fe-sulfides. Oxidation byproducts were in turn used by iron-metabolizing bacteria, forming large crusts of cell-entombing Fe-oxyhydroxides. The overall process, interpreted as the result of the stimulation at depth of specific activities, converted submicrometric Fe-sulfides into compact microbially-induced mineralizations, up to hundreds of micrometers in size, likely filling the basalt porosity, with a deleterious impact on the injectivity.