This study aimed to compare three approaches for predicting the service life of full-scale GAC adsorbers for the removal of micropollutants. The approaches included (i) rapid small-scale column tests (RSSCTs), (ii) two pilot-scale sampling approaches, and (iii) predictive correlations that consider micropollutant properties and background water matrix characteristics. The RSSCT could predict full-scale performance only if a micropollutant-specific fouling index was applied. At the pilot-scale, water samples were collected (1) over time at the top sampling point only (empty bed contact time (EBCT) of 1 minute) to minimize time to breakthrough (method 1) and (2) at different column depths at a single time point (method 2). Breakthrough curves obtained with method 2 more closely matched those obtained at the full-scale. In addition, method 2 is more convenient since it requires only one sampling campaign. Method 2 was used as a prognostic tool to predict breakthrough curves for micropollutants without full-scale data and a comparison with an existing prediction model gave satisfactory results for 6 out of 13 compounds.