Copper, nickel and lead boosted-discharge hollow cathode lamps, run at recommended currents, have been compared with a high-powered inductively coupled plasma (ICP) as excitation sources in atomic fluorescence spectrometry (AFS). A similar comparison was made with a copper lamp run at higher currents. It was found that for lead and nickel, the fluorescence spectra differed in the relative intensities of the transitions observed with the two sources. No evidence was found for a difference in radiances between the two sources when the lamp was overrun. Although the lamps gave rise to lower blank standard deviation values, detection limits were worse because of poorer sensitivity due to the inability of the circular source to illuminate the required atom cell volume in the atomiser. It was concluded that the ICP was the better source, when the criterion is detection limits, but the lamps may be more convenient in some circumstances.