In order to evaluate an enzyme-linked immunosorbent assay(ELISA) for practical use in detecting aflatoxin $B_1(AFB_1)$ from cereals, we compared $AFB_1$ concentrations of samples contaminated artificially or naturally that were quantitated by the ELISA with those spiked or quantitated by HPLC. Cotton seed meals(19 items), rape seed meals(ll), soybean meals(9), and corns(3) imported from foreign countries were used as sample cereals. The standard curves of each cereal class showed that 1-100 ng/g of $AFB_1$ from cereals could be assayed by the ELISA. When artificially contaminated cereals were assayed by ELISA, the average recovery of AFB! from samples spiked to 3 ng/g and more was 138%(68-193%), although that spiked to 1 ng/g was somewhat high(268%). The average C.V. of recovery was 7.0%(0-22.2%). When naturally contaminated cereals were assayed, the concentrations of $AFB_1$ below 10 ng/g especially from rape seed meals quantitated by ELISA were much lower than those determined by HPLC. However, the concentrations of 10 ng/g and more from samples, except a few extraordinary samples. quantitated by ELISA were similar to those determined by HPLC, especially in case of cotton seed meals whose average recovery (ELISA/HPLC) was 153%. In conclusion, the ELISA was elucidated such as a practical tool to detect $AFB_1$ of 10 ng/g and more from cereals.