A major advantage in using sequence stratigraphy is the ability to predict facies changes in sequences using the systems tracts approach. Glacial eustasy, however, imposes unique constraints on the facies distribution within and between sequences that challenge the predictive utility of sequence-stratigraphic models. The asymmetric rates of ice volume fluctuation result in a sawtooth eustatic curve and the formation of rapid transgressions and slow regressions. When subsidence rates are high, the augmented rate of formation of accommodation space allows the deposition and preservation of complete sequences in which there may be a repeated vertical succession of facies. A low subsidence rate, combined with variations in the rates of eustasy and sedimentation, results in unpredictable lateral and vertical facies patterns.

Data from the upper Pennsylvanian (Westphalian D to Stephanian) Conemaugh Group in the distal foreland of the Appalachian basin illustrate this problem. Within sequences, even the limited predictive ability associated with process facies models in the highstand systems tract is compromised by the temporal variations in the eustatic component and the presence of incised channels and valleys formed during late highstand and lowstand times. Furthermore, temporal variations in the driving functions mean that the presence of a particular facies in a systems tract of one sequence (e.g., coal) cannot be used to predict its presence or absence in a similar tract of an adjacent sequence. The result is that there are stratigraphic intervals, such as the Conemaugh Group, in which the detailed lithostratigraphic data from core or outcrop cannot be confidently extended beyond a few tens of meters within a sequence and not at all between sequences.