The mean at marriage is the most prevalent indicator for timing of marriage i demographic and social studies. Since calculation of this indicator requires information on the number of marriages and age at marriage, in most cases, the researcher is forced to use an indirect technique which is called the Singulate Mean Age at Marriage (SMAM). Nevertheless, the efficiency of SMAM as an estimate for mean age at marriage, is depends upon complicated assumptions regarding the persistence of marriage patterns. This paper uses census data, along with direct and indirect techniques in order to reject the assumption of persistence of marriage patterns among 1980s and 1990s marriage cohort women. It will then indicate that of SMAM not only has a low capability for estimating mean age at marriage but it also has a high degree of bias in a population with changing marriage patterns.