Unification of calibration curves for analysing micro-amounts of elements (N,S,C,Mo,Si,Ni,Mn,Cu) in tool steels and high speed steels was investigated. Iron contents in them vary from ca. 65 to 95%, so that classified calibration curves or correction of amounts of iron are necessary in the case of the internal standard method, while satisfied results were obtained by the fixed time integration method.For correction of interfering elements, iron base binary alloys were used for tool steels and multiple alloys were needed for high speed steels. Further, in order to decide coefficients of correction for high speed steels, the use of multiple regression analysis was investigated, using the calibration curves for only tool steels corrected with binary alloys, high speed steels were determined and the differences between the analytical values and the standared ones were used for calculation. Consequently, coefficients of correction were decided by using binary alloys for tool steels and by multiple regression analysis for high speed steels.The relation between sensitivities of analysis and coefficients of correction was also examined. As sensitivities the gradients of the calibration curves were chosen, and when the calibration curves were not straight, the gradients of regression lines in the trace region were used. The relation of hyperbola was observed between them; the larger the sensitivities the smaller became the change of coefficients of correction, and the lesser the sensitivities the smaller became the number of interfering elements.In tool steels, emission intensities were not affected by amounts of carbon from 0.2 to 1.4%, while they decreased in larger amounts of carbon (ca. 1.3%) in high speed steels.Tool steels and high speed steels were analysed with unified calibration curves. According to the results of the F-test, it was not concluded that there were differences in variance between tool steels and high speed steels and the accuracies of analysis were sufficiently high.