This paper examines the effect of earnings lags on estimating and comparing the extent of conservatism based on Basu's (1997) asymmetric timeliness concept. Based on earnings lags and earnings asymmetry, as well as the interaction between these two; we assert that any comparisons of the extent of conservatism should take the magnitude of both concurrent and cumulative asymmetric timeliness into consideration. The length of recognition lag should also be considered. In the present study, we construct a multi-period model to capture lags in earnings response to good news, as well as to bad news. We predict that positive asymmetric timeliness exists in the current period and with short lags. We also predict that asymmetric timeliness turns negative as lags increase. Using a sample of 74,550 U.S. non-financial and non-utility firm-years, empirical results are consistent with our predictions. Results are also robust to several changes in model and sample specifications.