Assuming that cells grow exponentially in length, the cycle of cells in chemostat cultures was analyzed parametrically by the age-size method (James, T. W. et al., Exp. Cell Res., 94, 267-276, 1975), which fitted well to the raw data of the length distribution. For simulation, we used the cell length and coefficient of variation at division from empirical data. The cell length and coefficient of variation at birth, the specific growth rate in extension, and the proportion of growing cells were estimated for simulation. We found that the stop-grow point of a cycle depended on the culture conditions: cells grown at D = 0.247 h-' in the chemostat culture had the stop-grow point at 0.72 of the cycle, similar to cells growing exponentially in the batch culture. Cells grown at D 0.126 h-' and D = 0.062 h-' in the chemostat cultures had the stop-grow points at 0.85 and 0.97 of the cycles, respectively. There was a sub-population of nongrowing cells estimated.Exponentially growing cells of the fission yeast, Schizosaccharomyces pombe, show a simple mode of growth: for the first 75% of the cell cycle, the cell grows in length and during the last 25% of the cycle there is little change in cell length (or volume, the constant-length or volume stage; 13,14). Twenty percent of the cells are in the constant-length stage, calculated by an ideal age-distribution curve (12). Accordingly, it is predicted that the length or volume distribution curve of cells should show a shoulder around the division length (length at division). There, however, is no such shoulder in volume distributions of steady-state cells in a chemostat culture (8), or cells growing exponentially on an agar-plate (5) in which the volume distributions were obtained by a Coulter Counter. In this study, we