We previously demonstrated a negative correlation between blood lead level and motor nerve conduction velocity in 202 asymptomatic 5 to 9-year-old children living near a lead smelter in Idaho. Blood lead levels ranged from 13 to 97 micrograms/dL. To determine whether a threshold exists between blood lead level and maximal motor nerve conduction velocity, we conducted three regression analyses on these data: a "hockey stick" regression, a logistic regression, and a quadratic regression. We found evidence for a threshold in all three analyses: at a blood level of 30 micrograms/dL in the "hockey stick" regression, at 20 micrograms/dL in the logistic, and at 25 to 30 micrograms/dL in the quadratic. Neither age, sex, socioeconomic status, nor duration of residence near the smelter significantly modified the relationship. These analyses confirm that asymptomatic increased lead absorption causes slowing of nerve conduction, but they also indicate that measurement of maximal motor nerve conduction velocity is an insensitive screen for low-level lead toxicity.