Seasonal cointegration generalizes the idea of cointegration to processes with unit roots at frequencies different from 0. Here, “common seasonals,” also a dual notion of common trends, is adopted for the seasonal case. The features are demonstrated in exemplary models for German and U.K. data. An evaluation of the predictive value of accounting for seasonal cointegration shows that seasonal cointegration may be difficult to exploit to improve predictive accuracy even in cases where seasonal non-cointegration is clearly rejected on statistical grounds. The findings from the real-world examples are corroborated by Monte Carlo simulation.