The time dependence of particle size (e.g., mean volume) in normal grain growth, bubble growth, and late‐stage coarsening is deduced from a statistical self‐similarity (SSS) hypothesis, according to which consecutive configurations of the system in the self‐similar mode are geometrically similar in a statistical sense, and from the scaling characteristic ofv˙, the rate of change of the volume of a given particle. It is shown that ifv˙ scales asv¯&agr;under uniform magnification, wherev¯ is the mean particle volume and &agr; a constant depending on the controlling kinetics and geometry of the system, thenv¯1−&agr;is a linear function of the time. Values of &agr; are obtained for a number of cases. The treatment is free of many of the approximations and geometrical simplifications used in most theories. Evidence for the SSS hypothesis is discussed. Special new results are obtained for two‐dimensional bubble and idealized grain growth that should be testable by computer simulation or by observation of a well‐behaved soap‐bubble array.