(1) The point-of-departure shall be the no-observed-effect level (NOEL), expressed as a dose in milligrams of substance per kilogram of body weight per day. Where a valid NOEL is not available, a lowest-observed-effect level (LOEL) may be used.

(2) If neither a NOEL or a LOEL are available, an alternative point-of-departure, e.g., the 95 percent lower confidence limit on the dose associated with a specified percentage of excess risk (e.g., 10 percent) for a nononcogenic effect adjusted for background risk, may be used. The alternative point-of-departure shall be estimated using one of the mathematical models that are appropriate for analysis of dichotomous or continuous dose-response data (e.g., power, polynomial, or linear), and shall be the model that best describes the dose-response data within the range of experimental observation. Statistical measures, including the Chi-squared goodness-of-fit test, shall be used to determine which model best describes the data.

(1) LOEL to NOEL extrapolation (where necessary, to account for uncertainty where extrapolating from a LOEL to a NOEL);

(2) subchronic to chronic extrapolation (where necessary, to account for uncertainty where extrapolating from a less-than-chronic study NOEL (or LOEL, or other point-of-departure) to a chronic NOEL, LOEL, or other point-of-departure;

(3) animal to human extrapolation (where necessary, to account for uncertainty where extrapolating from experimental animals to humans);

(4) inter-human variability (where necessary, to account for variation in sensitivity among the human population, including special consideration of the potential sensitivity of children); and

(5) data gaps (where necessary, to account for areas of scientific uncertainty in the toxicological database).