Fundamental aspects of various noise sources in flameless atomic absorption have been considered and the measurement system appeared to be shot noise limited. The influence of time constants of the measurement system, modulation frequency of the primary source, and heating rate of the sample cell on both noise level and signal-to-noise ratio have been experimentally investigated. Furthermore, the effect of the bandwidth of the amplifier-recorder system on analytical parameters, such as detection limit, sensitivity, and linear working range of the calibration curves, has been studied. Detection limits for undistorted signals, optimum signal-to-noise ratio, and highly distorted signals are reported.