@ mitch: The 3458A also has the OCOMP feature in the OHM modus (automatic substraction of thermal voltages). Disadvantage is the temperature sensitive, unstable OHM reference, and the high timely drift. A well ventilated cellar improves thingshat a lot .. I managed to reach there sub ppm OHM transfer stability (~ 0.2 ppm during 1h), although this is not specified at all. The 1nV resolution is superior on the 2002; the 3458A delivers 10nV only via Display (perhaps more and stable resolution via GPIB; I didn't yet understand completely the comparison document from Yongbin Leng). I encountered big programming trouble on the ancient Kethley 195A and 199A DMMs, should be gone on the 2002, but never saw that on any HP boxes. Latter ones firmwares are more stable than anything else to my opinion. The biggest advantage of the 3458A is the artifact calibration (complete AUTOCAL) with only two different sources. Didn't check, if the Keithley also got this feature, as a better differential linearity than 0.2ppm should be required, like in the 3458A. Frequent AUTOCALl helps to mitigate thermal instabilities, apart from its OHM cal drift. Frank