The values of dose and signal-to-noise ratio (SNR) for many techniques and breast thicknesses were computed and compared with a reference technique and breast thickness to provide a valid basis upon which to select a screening technique for screen-film mammography. The reference consisted of a Min-R/OM1 combination (or its equivalent) exposed through a 4.5-cm thick breast via 28 kV and a molybdenum target-beryllium window tube with a 0.03-mm Mo filter. Radiographs of an improved breast phantom were used to relate computed relative SNR values of techniques to diameters of the smallest calcific and soft-tissue objects demonstrated in mammograms. Without use of a grid, four Mo target/filtration combinations yielded similar computed dose and SNR levels, as did a tungsten target tube operated at 23 kV without a filter. Operation of the latter tube at 27.5 kV with a 0.051-mm rhodium filter should reduce dose by half but also SNR by 22%. However, such operation with a grid should greatly improve imaging moderate to large breasts without increasing dose over reference values.