Abstract: Context: The winter seeing at Concordia is essentially bimodal, excellent or quite poor, with relative proportions that depend on altitude above the snow surface. This paper studies the temporal behavior of the good seeing sequences. Aims: An efficient exploitation of extremely good seeing with an adaptive optics system needs long integrations. It is then important to explore the temporal distribution of the fraction of time providing excellent seeing. Methods: Temporal windows of good seeing are created by a simple binary process. Good or bad. Their autocorrelations are corrected for those of the existing data sets, since these are not continuous, being often interrupted by technical problems in addition to the adverse weather gaps. At the end these corrected autocorrelations provide the typical duration of good seeing sequences. This study has to be a little detailed as its results depend on the season, summer or winter. Results: Using a threshold of 0.5 arcsec to define the “good seeing”, three characteristic numbers are found to describe the temporal evolution of the good seeing windows. The first number is the mean duration of an uninterrupted good seeing sequence: it is $\tau0=7.5$ hours at 8 m above the ground (15 hours at 20 m). These sequences are randomly distributed in time, with a negative exponential law of damping time $\tau1=29$ hours (at elevation 8 m and 20 m). The third number is the mean time between two 29 hours episodes. It is T=10 days at 8 m high (5 days at 20 m).