Recent papers find that earnings volatility is again on the rise (Dynan et al. 2008, and Shin and Solon 2011). Using household survey data—the matched Current Population Surveys and Survey of Income and Program Participation—and the newly available Longitudinal Employment and Household Dynamics administrative dataset, we find that earnings volatility was remarkably stable in the 1990s and through the mid 2000s. This evidence is in contrast to that from the Panel Study of Income Dynamics (PSID) which registers a sharp increase in the early 2000s. We investigate whether adjusting measures based on our sources to more closely match the characteristics of the PSID can reconcile this divergence in trends, but do not find a clear explanation for the divergence. We also find little evidence of a rise over this period in the components of volatility: volatility among job leavers, volatility among job stayers, and the fraction of workers who are job leavers.