The sensory irritant properties of ozone have been considered to be responsible for symptoms that occur in humans after exposure. This assumption has not been studied explicitly. One way to assess the aversive properties of airborne irritants is to give the exposed individual an opportunity to control the duration of exposure, i.e., escape from the irritant. Mice were trained to turn off 1000-ppm ammonia, a concentration that, in humans, is irritating to the upper airways. Each mouse could terminate irritant delivery for 1 min by inserting its nose five times into one of two conical response sensors. Daily exposure was limited to a maximum of twenty-five 1-min exposures, every other minute. After the determination of ammonia concentration-effect curves, ozone was substituted for ammonia. Ozone exposures were alternated every other day with ammonia as a control for any changes that might occur as a result of repeated ozone exposure. Ozone reliably maintained escape behavior. Additional mice with no history of ammonia termination were trained to terminate ozone exposure, indicating that the aversive properties of ozone were not dependent on previous experience with other airborne irritants. As the concentration of ozone increased from 0.25 to 24 ppm, the number of escape responses increased, and the duration of ozone tolerated decreased. Ozone concentrations of 0.5 ppm or greater were significantly more aversive than control.