Cell-sediment separation methods can potentially enable determination of the elemental composition of microbial communities by removing the sediment elemental contribution from bulk samples. We demonstrate that a separation method can be applied to determine the composition of prokaryotic cells. The method uses chemical and physical means to extract cells from benthic sediments and mats. Recovery yields were between 5% and 40%, as determined from cell counts. The method conserves cellular element contents to within 30% or better, as assessed by comparing C, N, P, Mg, Al, Ca, Ti, Mn, Fe, Ni, Cu, Zn, and Mo contents in Escherichia coli. Contamination by C, N, and P from chemicals used during the procedure was negligible. Na and K were not conserved, being likely exchanged through the cell membrane as cations during separation. V, Cr, and Co abundances could not be determined due to large (>100%) measurement uncertainties. We applied this method to measure elemental contents in extremophilic communities of Yellowstone National Park hot springs. The method was generally successful at separating cells from sediment, but does not discriminate between cells and detrital biological or noncellular material of similar density. This resulted in Al, Ti, Mn, and Fe contamination, which can be tracked using proxies such as metal:Al ratios. With these caveats, we present the first measurements, to our knowledge, of the elemental abundances of a chemosynthetic community. The communities have C:N ratios typical of aquatic microorganisms, are low in P, and their metal abundances vary between hot springs by orders of magnitude.