The antibacterial activity of silver has long been known and has found a variety of applications because its toxicity to human cells is considerably lower than to bacteria. The most widely documented uses are prophylactic treatment of burns and water disinfection. However, the mechanisms by which silver kills cells are not known. Information on resistance mechanisms is apparently contradictory and even the chemistry of Ag + in such systems is poorly understood. Silver binds to many cellular components, with membrane components probably being more important than nucleic acids. It is difficult to know whether strong binding reflects toxicity or detoxification: some sensitive bacterial strains have been reported as accumulating more silver than the corresponding resistant strain, in others the reverse apparently occurs. In several cases resistance has been shown to be plasmid mediated. The plasmids are reported as difficult to transfer, and can also be difficult to maintain, as we too have found. Attempts to find biochemical differences between resistant and sensitive strains have met with limited success: differences are subtle, such as increased cell surface hydrophobicity in a resistant Escherichia coli. Some of the problems are due to defining conditions in which resistance can be observed. Silver(I) has been shown to bind to components of cell culture media, and the presence of chloride is necessary to demonstrate resistance. The form of silver used must also be considered. This is usually water soluble AgNO 3 , which readily precipitates as AgCl. The clinically preferred compound is the highly insoluble silver sulfadiazine, which does not cause hypochloraemia in burns. It has been suggested that resistant bacteria are those unable to bind Ag + more tightly than does chloride. It may be that certain forms of insoluble silver are taken up by cells, as has been found for nickel. Under our experimental conditions, silver complexed by certain ligands is more cytotoxic than AgNO 3 , yet with related ligands is considerably less toxic. There is evidently a subtle interplay of solubility and stability which should reward further investigation.