Written 27 July 2026 by Dr P, a counselling psychologist and EMDR practitioner
This is not training and it is not the protocol. It is what one clinician does between "start the set" and "write it up", which is the bit that never seems to get covered anywhere, presumably because it is unglamorous. If you have not done EMDR training, this page will not help you and you should not be running EMDR anyway. If you have, you may recognise the problem.
The short version. Sets are as long as they need to be and no protocol will give you a number that survives contact with an actual client. Time them anyway, because knowing what you did is worth something later. Write down one word per set, not a paragraph. Take SUD when the channel closes, not on a schedule. And leave yourself something usable when processing is incomplete, because next week you will have forgotten.
Nothing in EMDR requires a stopwatch. Therapists ran it perfectly well for years by counting passes and paying attention. Plenty still do.
The case for timing is quieter than that. During processing you are holding several things at once: where the client is, what just shifted, which set you are on, whether the last one moved anything. None of it is hard. All of it is a small tax on attention, payable at precisely the moment you would rather be paying attention to a person.
A timer does not make you a better clinician. It removes one of the things competing for the part of you that is listening, which is not the same claim but is the useful one.
The second reason is the record. Six sets in, "how long have we been going" is a question you will answer badly from memory and "was that set longer than the last one" is a question you will not answer at all.
The honest answer is that it varies and that anyone giving you a fixed number is selling something.
The conventional starting point in desensitisation is somewhere in the region of twenty four to thirty passes of bilateral stimulation, a pass being one full left to right cycle. In practice that lands somewhere around twenty to forty seconds depending on the speed you are running. Treat that as a place to begin rather than a target to hit.
What actually decides the length is what is happening in front of you:
Which is a long way of saying that a target duration is a prompt, not an instruction. If you use one, it should be something you can ignore without the app complaining at you.
Almost everyone starts by trying to write too much and gives up by set four. The trick is deciding in advance that a set gets one word and letting that be enough.
What is worth capturing:
What is not worth capturing mid-session: a narrative. You are not writing the notes yet, you are leaving yourself the raw material to write them from afterwards. The distinction matters because one of them can be done in two seconds without looking away and the other cannot.
The standard clusters are worth having somewhere you can glance at, not because they are hard to remember in the abstract but because they evaporate the moment you are actually in it:
The fifth is arguable and some frameworks fold it elsewhere. I keep it because it comes up.
The temptation is to take a SUD after every set, because it feels rigorous. It is not rigorous, it is intrusive and it teaches the client to monitor themselves rather than notice what comes.
Reasonable moments:
VoC belongs in installation and not before it. Asking how true a positive belief feels while somebody is still in the middle of the distressing material is a question with no useful answer and they will give you one anyway, which is worse.
Everything above is about the session you are in. This is about the next one and it is the part I got wrong for years.
When a session ends before something has resolved, which plenty do, the following week opens with archaeology. Where did we get to. How long were the sets running by the end. What came up in the last one. What was the SUD when we stopped. I was reconstructing all of that from memory and a few scattered notes, losing the first ten minutes of a session to it.
What is worth leaving yourself and it is not much:
Write it at the end of the session, not the end of the day. The gap between those two is where the useful detail goes to die.
Worth saying plainly, because a timer makes it easier to notice the clock and harder to notice the person. Running out of time is not a reason to stop processing abruptly. Leave enough of the session to close down properly: acknowledge the work, return to something settled, contain what is unfinished, then say out loud that it is unfinished and you will pick it up.
Whatever container or safe place work you use, use it. The client leaves your room and goes back to a life that will not pause politely while they finish processing.
I built EMDR Timer for exactly this, because nothing I could find did the small boring parts without also wanting to be my clinical record system. It times sets, counts them, takes a word and a rating per set and hands me a log I can paste into my notes. It is free, it works offline and the session data never leaves the device it was typed on.
You do not need it. A stopwatch and a notebook have worked for decades and will keep working. But if the tracking is costing you attention you would rather spend on the person in the room, that is the problem it was built to solve.