Two experiments using cylindrical samples of a dolomite-quartz rock were carried out in a conventional hydrothermal apparatus for the forward reaction: 1 dolomite + 2 quartz = 1 diopside + 2 CO2, in order to compare the mechanism and the kinetics with results from experiments using mineral powders of dolomite and quartz at the same P-T-X conditions. Experimental conditions were as follows: total pressure 500 MPa; temperature 680° C (overstepping 65° C); CO2 content of the fluid phase, consisting of carbon dioxide and water, was nearly 90 mol%; the fluid/rock ratio was 1:37, and the H2O/rock ratio was about 1:740; run duration was 92 days. Scanning electron microscope (SEM) examination of a polished axial section of the rock cylinders after the run, using back-scattered electrons (BSE), shows that the reaction produced corona textures. The diopside crystals nucleate and grow exclusively on dolomite surfaces adjacent to quartz grains, i.e. in regions where there is intimate contact between the reactants. The dolomite matrix, in contrast, is diopside free. A concept of microsystems is used to compare directly the rock cylinder results with those from runs done with mineral powders. The microsystems, which consist of quartz, dolomite and diopside, are connected by the intergranular space which is filled by the fluid phase. The SEM analysis of the rock cylinders indicates a dissolution-crystallization mechanism operating in the microsystems; this is consistent with the results of experiments using dolomite quartz powders (Luttge et al. 1989). It can be demonstrated that reaction kinetics in mineral powder runs are interface controlled as long as the newly formed diopside crystals do not cover the dolomite surfaces completely (Luttge and Metz 1991 c). This result is applicable to each microsystem of the rock cylinder, since the reaction mechanism and the resulting textures are the same in both kinds of experiments. The reaction is much slower outside the microsystems, i.e. in the dolomite matrix but in the close vicinity of the quartz grains. At these places, the reaction is controlled by the transport of Si-species in the CO2-rich fluid phase filling the intergranular space. The reaction is absent in quartz-free regions of the dolomite matrix. Calculations and measurements of the extent of reaction progress in both kinds of experiments give results of the same order of magnitude: the conversion, and therefore the reaction rate, differs by less than a factor of two. The conclusion is that there are no differences, in principle, concerning mechanisms, rate controls, rates, and resulting textures between rock cylinder experiments, and mineral powder experiments.