Hydrogeochemical models for the prediction of drainage quality from full-scale mine waste-rock piles are often parameterized using data from small-scale laboratory or field experiments of short duration. Yet, many model parameters and processes (e.g., sulfide-oxidation rates) vary strongly with the spatiotemporal dimensions of the experiment: the “upscaling” of prediction models remains a critical challenge for mine-waste management worldwide. Here, we investigate scale dependence in laboratory and field experiments that spanned orders-of-magnitude in size (i.e. 2 kg to 100,000 kg) at the Antamina mine in Peru. Normalized drainage mass loading rates systematically decreased with increasing scale, irrespective of waste-rock type. A process-based reactive-transport model was used to simulate observed rates and reproduce the geochemical composition of drainage across scales. Long-term trends in drainage quality could be quantitatively reproduced when the model was parameterized with mostly scale- and experiment-specific measured bulk properties or literature values, leaving geochemical rate coefficients the sole calibrated model parameters. Analysis of these fitted parameters revealed that the scale dependence of geochemical rates was largely explained by reactive mineral surface area. This work demonstrates that practical drainage quality predictions for full-scale waste-rock piles can be established from readily available bulk parameters determined at multiple scales.