A cylindrical sample of a tremolite-dolomite marble was used as starting material for a long-term experiment of 184 days at 581 °C and a CO2–H2O-fluid pressure of 100 MPa. The fluid, which was always present in excess, had a composition in the range of XCO2 = 0.21−0.37. During the run forsterite and magnesian calcite formed both within the rock sample as well as on its cylindrical surface. In areas where a tremolite crystal was exposed on the sample surface a 400–500 μm thick forsterite – magnesian calcite reaction rim evolved with a composition of 58 ± 2 mol% forsterite and 42 ± 2 mol% calcite. The dolomite areas of the surface of the rock cylinder are similarly covered by a reaction rim of forsterite plus magnesian calcite. This ca. 300 μm thick rim, however, is composed of 28 ± 5 mol% of forsterite and 72 ± 5 mol% of calcite. From the different compositions of the reaction rims we conclude that the well-known overall reaction: 1 tremolite + 11 dolomite ⇔ 8 forsterite + 13 calcite + 9 CO2 + 1 H2O occurs via the following two partial reactions: (1) 1 tremolite + 2 CO2 + 21 H2O ⇔ 2.5 forsterite + 2 calcite + 5.5 Si(OH)4 · 2 H2O and (2) 11 dolomite + 5.5 Si(OH)4 · 2 H2O ⇔ 5.5 forsterite +11 calcite + 11 CO2 + 22 H2O. All three equations are slightly simplified, because they assume the formation of pure instead of magnesian calcite. The derived partial reactions, which occur as expected via dissolution-transport-precipitation mechanisms, are sequential reactions; they are linked by the production, diffusion, and consumption of the component Si(OH)4 · 2 H2O. In contrast to the two kinds of rims on the rock cylinder’s surface, only one combined forsterite – magnesian calcite reaction rim was formed between tremolite and dolomite in the interior of the sample. The formation of this reaction rim can be explained by a combination of the two partial reactions given above. For the formation of forsterite plus magnesian calcite on the surface of the rock cylinder we come from an interpretation of the experimental results to the conclusion that the diffusion of the component Si(OH)4 · 2 H2O through the forsterite – calcite reaction rim that developed on dolomite is the rate-controlling step in the sequence of the two partial reactions given above. For the formation of forsterite plus calcite within the rock sample it is not possible to derive a rate-controlling step for the overall reaction, because of different diffusion paths of the reaction fluid and the components dissolved in it. The observed partial reactions are crucial for deciphering the formation of the various textures of forsterite – calcite –dolomite (−tremolite) assemblages developed in metamorphic siliceous dolomites. The magnesian calcites of the reaction rims have different compositions, which all deviate considerably from the equilibrium composition according to the calcite-dolomite solid solution. This result sheds light on the difficulty to attain equilibrium between the reactant dolomite and the product magnesian calcite. The consequences of the observed disequilibrium for the calcite–dolomite geothermometry are discussed.