Bacteria tightly regulate and coordinate the various events in their cell cycles to duplicate themselves accurately and to control their cell sizes. Growth of Escherichia coli, in particular, follows a relation known as Schaechter's growth law. This law says that the average cell volume scales exponentially with growth rate, with a scaling exponent equal to the time from initiation of a round of DNA replication to the cell division at which the corresponding sister chromosomes segregate. Here, we sought to test the robustness of the growth law to systematic perturbations in cell dimensions achieved by varying the expression levels of mreB and ftsZ. We found that decreasing the mreB level resulted in increased cell width, with little change in cell length, whereas decreasing the ftsZ level resulted in increased cell length. Furthermore, the time from replication termination to cell division increased with the perturbed dimension in both cases. Moreover, the growth law remained valid over a range of growth conditions and dimension perturbations. The growth law can be quantitatively interpreted as a consequence of a tight coupling of cell division to replication initiation. Thus, its robustness to perturbations in cell dimensions strongly supports models in which the timing of replication initiation governs that of cell division, and cell volume is the key phenomenological variable governing the timing of replication initiation. These conclusions are discussed in the context of our recently proposed "adder-per-origin" model, in which cells add a constant volume per origin between initiations and divide a constant time after initiation.acteria can regulate tightly and coordinate the various events in their cell cycles to accurately duplicate their genomes and to homeostatically regulate their cell sizes. This is a particular challenge under fast growth conditions where cells are undergoing multiple concurrent rounds of DNA replication. Despite much progress, we still have an incomplete understanding of the processes that coordinate DNA replication, cell growth, and cell division. This lack of understanding is manifested, for instance, in discrepancies among various recent studies that propose different models for control of cell division in the bacterium Escherichia coli.One class of models suggests that cell division is triggered by the accumulation of a constant size (e.g., volume, length, or surface area) between birth and division (1-3). Such models are supported by experiments measuring correlations between cell size at birth and cell size at division, which showed that, when averaged over all cells of a given birth size vB , cell size at division vD approximately followswhere the constant v0 sets the average cell size at birth. This is known as the "incremental" or "adder" model, and cells following this behavior are said to exhibit "adder correlations" (1-7). Importantly, these models postulate that cell division is governed by a phenomenological size variable, with no explicit reference to DNA replication. A...