AbstractThe possibility of recrystallization is a long‐standing barrier to deciphering the genetic origin of dolomites. There is often uncertainty regarding whether or not characteristics of ancient dolomites are primary or the consequence of later recrystallization unrelated to the original dolomitization event. Results from 65 new high‐temperature dolomite synthesis experiments (1 m, 1·0 Mg/Ca ratio solutions at 218°C) demonstrate dolomite recrystallization affecting stoichiometry, cation ordering and nanometre‐scale surface texture. The data support a model of dolomitization that proceeds by a series of four unique phases of replacement and recrystallization, which occur by various dissolution–precipitation reactions. During the first phase (induction period), no dolomite forms despite favourable conditions. The second phase (replacement period) occurs when Ca‐rich dolomite products, with a low degree of cation ordering, rapidly replace calcite reactants. During the replacement period, dolomite stoichiometry and the degree of cation ordering remain constant, and all dolomite crystal surfaces are covered by nanometre‐scale growth mounds. The third phase (primary recrystallization period), which occurs in the experiments between 97% and 100% dolomite, is characterized by a reduced replacement rate but concurrent increases in dolomite stoichiometry and cation ordering. The end of the primary recrystallization period is marked by dolomite crystal growth surfaces that are covered by flat, laterally extensive layers. The fourth phase of the reaction (secondary recrystallization period) occurs when all calcite is consumed and is characterized by stoichiometric dolomite with layers as well as a continued increase in the degree of cation ordering with time. Inferences of recrystallization, in natural dolomite, based on cation order or stoichiometry of dolomite, usually depend on assumptions about the precursor dolomite subjected to recrystallization. If it is assumed that the experimental evidence presented here is applicable to natural, low‐temperature dolomites, then the presence of mounds is direct evidence of a lack of recrystallization and the presence of layers is direct evidence of recrystallization.