This question was originally asked several years ago but it is still the subject of debate.

The question is about connecting test instruments (up through 480 V) for taking measurements such as voltage, harmonics, transients etc.

The debate is whether to de-energize the circuit first, connect the test instrument and then re-energize the circuit vs. simply connecting the instrument.

Here is the question:

When using a test instrument for taking measurements on energized circuits , what method do you / would you use for making the connection?

De-energize, connect the instrument, re-energize
Connect the instrument while energized

99% of the time the answer is connect while energized; however, there are occasionally instances where I opt to de-energize due to perceived risk.

Since it isn't deenergized until it's been tested for absence of potential, there can be only one possible answer.

Stevenal is absolutely correct, in an absolute context. There may be an occasion where it is appropriate to de-energize first however, depending of course on what it is your are doing.
Bottom line: what you are doing (the testing actions and purpose)as well as the potential for instrument or system failure and consideration of the IE level are going to determine the safety actions you take.

It depends on where the connections can be made and what the voltage and incident energies are.

If the connection can be made downstream of HRC fuses, the incident energy would be fairly low. If there is comfortable access and lots of space, the risk of accidental contact and flashover would be low.

It also depends on the consequences of turning the power off rather than leaving the circuit energized (do you defeat safety systems?; interrupt medical equipment?, etc).

If it makes no difference to the electrical load, de-energize it. Otherwise, for low risk applications, I would plan the work and then do it energized. As the risk grows, you need to evaluate how important is the reading? Can it wait until the circuit can be de-energized?

There is a more fundamental question: What are you testing for?

If you are troubleshooting, voltage or current probably needs to be present.

If you are taking time recorded event such as power quality it is possible to connect the equipment cold. It has been my experience when connecting a PQ meter I will generally start at the panel main. The effort to get the utility to kill power is far greater than the time to simply connect the meter. If the service is large enough there most likely is not PPE available for arc flash. Shock is probably the bigger concern - which may lead to arc flash.

Connect while energized, typically need voltage present as part of the troubleshooting.

Now typically wear more PPE (i.e, voltage rated gloves, etc.) depending on risk level than in the past, which is a good thing.

Same answer here...situational dependent.

But more fundamentally, is PPE even required?

Example...using the insulated ends of a multimeter on a 120 V control circuit. They are already insulated with guards so it meets the definition of not being exposed so gloves should not be necessary. Similar arguments can and should be made with regard to capacitive voltage sensors on hot sticks. As to arc flash, if the tips are insulated sufficiently so that a minimum of metal is exposed and this definition depends on the amount of exposed metal and the distance between energized phases (exposed metal), there is no chance of arc flash in the first place when doing a risk analysis.

Moving on from voltage tests, tests with a current clamp can quite often be done again with no exposed conductors and no chance of initiating an arc flash in the first place. Whether gloves are required depends on the design of the clamps and position of the equipment, and what the voltages are involved. The tricky part comes in when placing them around wiring that may be suspected of being compromised where the insulation may fail and the act of putting the current clamp on causes an arc flash from disturbing the wiring. No chance of this on say solid bus bars but depending on age and condition, some cables may be a different matter. This is where eminent danger of failure definitions come into play. If we are definitely in a state where there is no exposed wiring and the insulation is not in danger of failing and causing exposure then by all means, current clamps are safe to use without arc flash hazards. The trouble here of course is that there is some measure of judgement call to be made and this depends on the operation at what level they allow a jdugment call to be made, if at all. The trouble of course with "no tolerance" policies is that they create all kinds of idiotic situations.

There may be an occasion where it is appropriate to de-energize first however, depending of course on what it is your are doing.