Conventional figures of merit such as limits of detection, signal to background ratio or repeatability, are used to determine the performance of pneumatic and ultrasonic sample introduction systems in an ultra-clean environment with an axially viewed inductively coupled plasma-atomic emission spectrometry and multichannel detection. We observed that the ultrasonic nebuliser offered a large improvement of signal intensity (10-133 greater) compared to a cyclone chamber coupled with a pneumatic Meinhard nebuliser. This improvement is associated with an average increase of signal to background ratio by a factor 86 and an average decrease of detection limits by a factor 6. The improvement factors generally depend on the element and for the same element on spectral lines. Typically, the observed values of detection limits in this work are lower than those published and obtained in non-ultra-clean conditions. The results emphasize that the environmental conditions of cleaning and analysis are essential to avoid and control cross contamination of the samples and hence to obtain low detection limits.