Seminal advances in single-particle analysis have led to countless applications of cryo- electron microscopy (cryo-EM) for the determination of molecular structure, such that limitations in sample preparation are now considered to be the major obstacle for routine application of this method. Problems with sample preparation can be thought of in terms of the vastly different behaviour of different molecular species when confined to a thin layer of ice (preferential orientation, dissociation), and the need to obtain consistently thin ice, where ‘thin’ refers to dimensions on the order of the molecule or complex under study. In contrast to previous studies showing ice thickness measurement ‘after the fact’, we considered the need to screen frozen-hydrated sample grids for selection of appropriate areas before data acquisition, thus enabling the inclusion of this procedure in a routine workflow. Screening was performed at the magnification used for selecting areas manually or via greyscale filtering. We used a Bioquantum energy filter in combination with a K2 direct electron detector (Gatan, Pleasanton, USA) to calculate the ice thickness for test specimens (apoferritin, green fluorescent protein) on gold Quantifoil-type sample grids (Plano GmbH, Wetzlar, Germany) and correlated this information with independent measurements based on ice tunnels and electron tomography. All methods showed broad agreement. As expected, measurement of tunnels made with an induced sample tilt became increasingly inaccurate as ice became thinner due to increasing measurement uncertainty. Like tunnelling, tomography is unsuitable for routine application but it served to calibrate and corroborate measurements made using the energy filter. Using this approach, we were able to combine the greyscale measurements provided by Thermo Fisher’s EPU software and our energy filter measurements to categorise the ice quality (thickness), and thereby prioritise areas for data acquisition.