The present study was initiated to examine the quantitative relationship between blood lead (Pb-B) and urinary δ δ δ δ δ-aminolevulinic acid (ALA-U) among Pb-exposed workers, and to find a threshold Pb-B level to induce an increase in ALA-U. For this purpose, pairs of venous blood and spot urine samples were collected from 8,274 men and 5,856 women (14,130 workers in total) who were occupationally exposed to inorganic lead. The blood and urine samples were analyzed for Pb-B and ALA-U by atomic absorption spectrometry and colorimetry, respectively, and the correlation between pairs of measures were subjected to statistical analysis. The assumption of the 3rd degree regression for correlation gave a substantially greater correlation coefficient (0.645 for men and 0.619 for women) than 1st or 2nd degree regression, whereas only very small improvement in the coefficient was achieved with 4th to 6th degree ones. Logarithmic conversion of the parameters was not effective in improving the correlation. The assumption of the 3rd degree regression followed by calculation of the local minimum gave 22, 29 and 23 µg/100 ml Pb-B for men, women, and men+women, respectively, as the threshold Pb-B to induce ALA-U increase. Pb-B to elevate ALA-U to the 95% upper normal limit (8 mg/l, common to men and women) was 62, 50 and 58 µg/100 ml for men, women and men+women, respectively. The validity of the 3rd degree regression assumption as a tool to calculate a threshold from experimental or epidemiological data is discussed.