The possibility of using more than one line for calibration in optical emission spectroscopy has been investigated. By means of principal component regression, sensitivity can be increased and part of noise can be removed. Thus, lower limits of detection can be obtained. The extent of improvement of sensitivity is considered theoretically and investigated by simulation. In the ideal case the sensitivity of the principal component regression for more than one line is calculated from the square root of the sum of squares of the sensitivities of the single lines. Practical applications in arc-optical emission spectroscopy and ICP-OES show that the theoretically expected improvements can almost be reached. The reduction in noise depends on the nature and source of noise.