Preconcentration of run-of-mine subgrade and marginal ore by screening has the potential to substantially improve the economics of mining, particularly for large, low-grade orebodies, and to reduce environmental impacts. It has recently been the subject of a major research effort by a joint public-private research body, CRCORE. Yet despite this, there appears to have been only one fully commercial installation, at Bougainville Copper Limited in the late 1980s. Two reasons for the limited application are that not all ores have suitable properties and the economics of all pre-concentration processes on low-grade materials are highly sensitive to the precise balance of mass and value recovery and to the pre-concentration process cost, including materials handling. These factors make it imperative to have good test data on the process response of feedstock across the orebody. Collecting that data presents major challenges, which differ for green and brownfields operations. This paper first discusses the approach to preliminary evaluation developed by CRCORE, which relies on testing of coarse rejects or drill chips available from exploration drilling programs. This method is an efficient first step in evaluating an orebody which is deemed geologically to be potentially amenable to screen upgrading, but it does have limitations and must be followed by larger scale work. A suite of methods have been developed by the authors as a result of test work since the 1990s on several projects, supplemented by published data, which aim to provide statistically robust data at lower cost than large scale bulk sampling tests. For greenfields projects, the upgrade response measured on material crushed to -53 mm has been found to correlate closely with ROM ore, with no scale-up required for material with upgrade responses in the range of commercial interest. Large diameter whole drill core, such as PQ, is expected to be suitable for this method. As drilling such core is costly, the aim should be to establish relationships between upgrade performance and proxy measurements such as multi-element chemical or hyperspectral analyses. This approach was successful at the Gramalote gold project in Colombia. The method is not applicable for ores with thick vein material comparable with the crush size, but such material is not suitable for upgrading by screening and would be eliminated by the preliminary testing. For brownfields projects where blasted material can be sampled, the upgrade response can be accurately measured by sampling only the finer fraction of the ore, selecting a top size which represents 30–50 per cent or more of the ROM size distribution. This is likely to be between 50 and 100 mm. The much finer size compared with the full ROM ore greatly reduces the sample mass required to achieve a given intrinsic sampling error and enables sample preparation to be undertaken with large-scale laboratory equipment rather than mobile crushing plant. The accuracy of the method can be improved by measuring the approximate mass of oversize rejected during sampling. Some ores do show a systematic deviation from the assumed model fit which results in this method slightly underestimating the upgrade response, and it would therefore still be recommended to take one or two full bulk samples to confirm results before a final investment decision.