Gold nanoparticles (AuNPs) and gold ion complexes have been investigated for their antibacterial activities. However, the majority of the reports failed to disclose the concentration of free Au(I) or Au(III) present in solutions of AuNPs or gold ion complexes. The inconsistency of antibacterial activity of AuNPs may be due to the effect of the presence of Au(III). Here we report the antibacterial activity of Au(I) and Au(III) to four different bacteria: one nonpathogenic bacterium: E. coli and three multidrug-resistant bacteria: E. coli, S. typhimurium DT104, and S. aureus. Au(I) and Au(III) as chloride are highly toxic to all the four bacteria, with IC50 of 0.35 – 0.49 µM for Au(III) and 0.27–0.52 µM for Au(I).The bacterial growth inhibition by both Au(I) and Au(III) increases with exposure time and is strongly affected by the use of buffers. The IC50 values for Au(I) and Au(III) in different buffers are HEPES (0.48 and 1.55 µM) > Trizma (0.41 and 0.57 µM) > PBS (0.14 and 0.06 µM). Bacterial growth inhibition by AuNPs is gradually reduced by centrifugation-resuspension to remove residual Au(III) ion present in the crude synthetic AuNPs. After 4 centrifugations-resuspensions, AuNPs become non-toxic. In addition, both Au(I) and Au(III) are cytotoxic to skin keratinocyte and blood lymphocyte cells. These results suggest that Au(I) and Au(III) in pure or complex forms may be explored as a method to treat drug-resistant bacteria, and the test of AuNPs toxicity must consider residual Au(III), exposure time, and the use of buffers.