Time series sometimes offer special features, such as the natural disaggregation of their components according to a hierarchical structure. Hierarchical forecasting methods can take advantage over such structure by considering the base forecasts reconciliation, producing results which are usually unbiased and more accurate than the ones provided by standard methods. In this work, we introduce the concept of robust estimation for hierarchical forecast reconciliation methods. We formalize two different robust approaches applied to data from a Labor Force Survey in Brazil. To demonstrate the potential and validity of the proposed approaches, we compare their performance with those from traditional and state-of-the-art methods. We provided an original contribution to the field of Hierarchical Time Series adopting the framework of robust estimation. From an empirical evaluation, the results reveal a partial dominance of the robust techniques over the optimal combination based on weighted least squares, bottom-up and top-down, indicating a potential field of study in the subject of Hierarchical Time Series.