Cobalt-57 flood sources are often used for system flood-field uniformity checks of scintillation camera performance. Such sources are now available in large sizes and with high activities. Uniformity images using new 57Co sources have shown artefacts which are not present in uniformity images obtained from technetium-99m flood sources of the same activity. The high-energy photons emitted by 57Co and cobalt isotope impurities appear to influence the images obtained. This was investigated in three cameras using three 57Co sources of different age. The flood-field images obtained with the 57Co sources showed non-uniformity patterns that were dependent on the age of the cobalt source, the distance of the source to the collimator, and the specific camera type. Quantification of the uniformity images reflected these findings. Energy spectra of a new 57Co source, obtained with an external 1024-channel analyser connected to the camera, showed a broad tail of high-energy photons above the 122-keV photopeak, due to Compton scatter and collimator penetration. This tail diminished with older sources and with increased source to collimator distance, indicating that in both situations fewer high-energy photons were being measured by the camera system. This tail of high-energy photons contributes to the total count rate, but because this is not obvious from the counts observed in the photopeak window, the camera can unsuspectingly be operated at too high a total countrate where pile-up effects become significant. Caution is therefore advised when using 57Co sources. Problems may be minimized by purchasing low-activity sources (certainly no more than 370 MBq), by placing the source at a distance from the collimator, e.g. 50 cm, by allowing a new source to decay, and by manufacturers producing 57Co sources without detectable radioactive contaminants.