Deos anyone have a "rule of thumb" about the correct area to noise threshold for analysing low levels of sample? We have the default setting at about 3000uV, but a colleague of ours says that we should have it at about 100.

Generally we would look at Signal - Noise using peak height, rather than area (since you can't generally measure the area of noise easily). For this, Limit of Detection is set at a S/N ratio of 3/1, i.e. where the peak height is 3 times the average noise around the peak of interest. Limit of Quantitation would generally be higher than this - we have a general LOQ level of 0.1% of main expected peak size.