Decline curve analysis is the most commonly used technique to estimate reserves from historical production data for evaluation of unconventional resources. Quantifying uncertainty of reserve estimates is an important issue in decline curve analysis, particularly for unconventional resources since forecasting future performance is particularly difficult in analysis of unconventional oil or gas wells. Probabilistic approaches are sometimes used to provide a distribution of reserve estimates with three confidence levels (P10, P50 and P90) and a corresponding 80% confidence interval to quantify uncertainties. Our investigation indicates that uncertainty is commonly underestimated in practice when using traditional statistical analyses. The challenge in probabilistic reserves estimation is not only how to appropriately characterize probabilistic properties of complex production data sets, but also how to determine and then improve the reliability of the uncertainty quantifications. In this paper, we present an advanced technique for probabilistic quantification of reserve estimates using decline curve analysis. We examine the reliability of uncertainty quantification of reserve estimates by analyzing actual oil and gas wells that have produced to near-abandonment conditions, and also show how uncertainty in reserves estimates changes with time as more data become available. We demonstrate that our method provides more reliable probabilistic reserves estimation than other methods proposed in the literature. These results have important impacts on economic risk analysis and on reservoir management.