Laser-ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) has become a powerful tool in geochemistry, but for some important elements it is prone to interferences. The detection capabilities of optical emission spectroscopy (OES) might be superior to quadrupole mass spectrometry (QMS) in certain cases. Experiments for Na, Mg, Si, K, Ca and Fe showed limits of detection (LOD) with QMS of 0.9, 0.3, 700, 7.1, 180, 22 mu g g(-1), respectively, obtained for continuous ablation of the Glass standard SRM 612 from NIST at a rate of approximate to 5.5 ng s(-1) in a 40 mu m ablation pit. A modern OES system linked to the same LA setup showed poorer LOD of one to two orders of magnitude for these elements except for Ca. This is mainly due to the better sensitivities of QMS resulting from higher signal-to-background ratios. Yet, from the recording of short transient signals by QMS generated by the ablation of similar to 500 pg of material using a single laser shot, 20-40 times higher LOD have to be accepted relative to continuous ablation. For menus exceeding approximate to 10 isotopes with dwell times above 10 ms, the scan speed of the sequential QMS filter relative to the signal duration may be insufficient for representative sampling of short (<approximate to 3 s) transient signals, leading to less reproducible results. Simultaneous signal recording by OES has the potential to eliminate this problem, if the LOD of OES are low enough to detect such signals. Unfortunately, the duty cycle of the OES system degraded significantly with increasing number of spectral lines, due to the slow data readout, from 35% (10 elements) to 12% measurement efficiency (18 elements). The example of natural fluid inclusions shows that, for multi-element menus and short transient signals, LA-ICP-QMS is currently the best method, until the sensitivity, dynamic range of the CCD and data readout speed of OES instruments are improved.