The paper addresses the problem of uncertainty in the greenhouse gas emission inventories, by proposing an alternative method for assessing uncertainty and its evolution over time. To estimate the inventory accuracy, the revisions published in consecutive years are used. These revisions are considered nonstationary time series. We describe evolution by time-dependent models, used to analyze data from the National Inventory Reports published annually up to 2015, for selected EU countries. We present a parametric model and a procedure for estimating parameters, along with the results obtained.