New human estimates for chlorine inhalation lethality as a function of exposure duration (for a healthy subpopulation and the general population) were derived via a review and statistical analysis of existing mammalian lethality data. Such estimates are needed to support risk assessments and casualty predictions involving chlorine airborne releases. At present, casualty predictions for such releases are at odds with what has been observed historically; the predicted downwind hazard area has often been much larger than what was actually observed. Either the present toxicity estimates are too low, the currently popular atmospheric transport and dispersion (ATD) models cannot adequately model chlorine releases, or both.

Median lethal dosage and quantal response data were found and analyzed for eight species (mouse, rat, guinea pig, rabbit, cat, dog, goat and sheep) and for durations from 8 to 235 minutes. The base 10 probit slope (concentration) was estimated via the weighted average of experimentally measured slopes in mammalian lethality studies. Resulting human lethality (military) estimates as a function of exposure duration were expressed via the toxic load model. General population estimates were derived from the military estimates using the mathematical method of Crosier (2007).

Previous human estimates were reviewed and one study identified as corresponding to the lower confidence limit for the new general population estimate. The impact of the new estimate was evaluated through a series of transport and dispersion modeling runs for the catastrophic accidental release of 50 tons of chlorine from a tanker car. The sensitivity of downwind hazard distances was also investigated as a function of median lethal toxic load, toxic load exponent and probit slope values.