This study was performed to determine whether ultrasound of frequency and average intensity that are used in therapy, could produce strand breaks in the DNA of suspended sheep peripheral leucocytes and mouse thymocytes. A clinical therapy applicator with a transducer of resonant frequency at 870 kHz was used for experiments. Strand breaks in DNA, measured by a fluorescence enhancement assay, were 2 initiated in sheep leucocytes at 1.2 W/cm and increased linearly with intensity to a maximum at 4 ..5 W/cm . No repair of breaks was observed following a 20 min incubation at 37 C. No strand breaks were detected in Swiss-Webster mouse thymocytes exposed to the same conditions, although breaks were detected in cells from a Senear strain. The thiol cysteamine, an intracellular scavenger of free radicals, protected cells from DNA damage while the disulfide cystamine, which cannot penetrate the cell, did not. High viscosity medium of methyl cellulose which prevented cavitation conditions which could give rise to free radicals, prevented DNA damage in leucocytes. Superoxide, *02, and hydroxyl radicals, OH', were produced in salt solutions under experimental conditions. The ultrasound doses for initiation and increase in production of OH' coincided with those for DNA strand breakage in cells. These results are consistent with a model in which free radicals formed in the extracellular medium during exposure to ultrasound can lead to strand breaks in DNA. if cells in vivo exist in media of low viscosity, it is possible that medical applications of ultrasound could produce damaged DNA.