Depends on the 'bandwidth' of the signal and the test equipment and the individual. Generally consider many non-medical measurements below 025mA suspect, as few design engineers make an effort to understand the various frequency components of the voltage and current leakage signals, and/or understand the effects to the signal of instrument impedances.

Have had several discussions with Mr.Pete Perkins (he sits on several STCs and has written several IEEE papers) on this subject, and we agree that the human body model referenced by these safety standards being used for the measurement network (IEC60990) have problems because the medical community does not understand physics. As the various standards committees continue to look at the body of work being done by bioengineers, will probably see better measurement methods being codified.

Human response and perception to electricity has significant variance and resultant effects/affects. My wife can detect less than 10uA at 200Hz/42V. My detection threshold is at least an order of magnitude greater, even at 50Hz.