Background: Optimisation of average glandular dose (AGD) for two-dimensional (2D) mammography is important, as imaging using ionizing radiation has the probability to induce cancer resulting from stochastic effects. This study aims to observe the effects of kVp, anode/filter material, and exposure mode on the dose and image quality of 2D mammography. Methods: This experimental study was conducted using full-field digital mammography. The entrance surface air kerma was determined using thermoluminescent dosimeter (TLD) 100H and ionization chamber (IC) on three types of Computerized Imaging Reference System (CIRS) phantom with 50/50, 30/70, and 20/80 breast glandularity, respectively, in the auto-time mode and auto-filter mode. The Euref protocol was used to calculate the AGD while the image quality was evaluated using contrast-to-noise ratio (CNR), figure of merit (FOM), and image quality figure (IQF). Results: It is shown that AGD  values in the auto-time mode did not decrease significantly with the increasing tube voltage of the silver filter (r = −0.187, P > 0.05) and  rhodium filter (r = −0.131, P > 0.05) for all the phantoms. The general linear model showed that AGD for all phantoms had a significant effect between different exposure factors [F (6,12.3) = 4.48 and mode of exposure F (1,86) = 4.17, P < 0.05, respectively] but there is no significant difference between the different anode/filter combination [F (1,4) = 0.571]. Conclusion: In summary, the 28, 29, and 31 kVp are the optimum kVp for 50%, 30%, and 20% breast glandularity, respectively. Besides the auto-filter mode is suitable for 50%, 30%, and 20% breast glandularity because it is automatic, faster, and may avoid error done by the operator.