Background : The photoperiod and temperature were ruled out as main factors of the annual menarche rhythm. Vacations associate to peaks and study periods to troughs of menarche. Some months show contradictory behavior according to their vacation-study condition. Aim: To test the hypothesis that menarche occurs mainly in relation to fiesta or celebration (birthdays, holidays, and vacations) expectancies. To refute the memory bias in recording the day at menarche. Subjects and methods: Girls from Santiago, Chile (3,957) and Medellin, Colombia (3,616) were asked about the date of birth and menarche. Results: In both cities, menarche occurred more commonly at birthday, Christmas-New Year, Day of the Independence, other holidays and days at vacation. The bias of memory was refuted because in both cities important holidays did not have excesses of menarche and some non-holidays had excesses; the distribution of menarches in the neighbor of peaks was often a left biased bell, with accessory peaks, and menarches/day over the expected mean (in agreement with the fiesta expectancy), and not a U shaped distribution (as expected by the memory bias); menarche distributed homogeneously on days of the week. Conclusions: The fiesta expectancy appears as a cause of the menarche rhythm. The bias of memory does not produce detectable effects. Evidence for recruitment of menarches and synergism between fiesta expectancy and vacations were found (Rev Méd Chile 2004; 132: 299-306).