Update: I have updated language using stop values, instead of EV values to prevent confusion from the absolute EV/LV standard. I hope this clears up the confusion. The results and the performance of the cameras remain unchanged. I welcome constructive criticism as it would be easy to retest. Thanks 
This is going to be a bit of a gear head, nerdy type test which I occasionally do when I am really intrigued with something about a camera or set of cameras. For the sake of all of our sanity, I will bold the important, take home messages so you can skim if you like.
The Canon 6D’s center focusing square touts a rather extraordinary low light focusing ability to – 3EV (EV meaning Exposure value, or stop of light), which means the 6D should focus well in darker situations. As a wedding and portrait photographer, I cannot emphasize how important it is to have a camera that can focus in low light, especially in a reasonably quick manner. On paper, the Nikon D600 is rated at -1 EV.
I conducted a series of tests this afternoon at my place, and it will take a minute to explain how this was done for any of this to make any sense. There may even be some flaws, but I believe these results are going to be pretty easy to duplicate under the same circumstances. This was not a test for focusing precision (whether it was tack sharp or not), this was a test of how fast these 4 cameras can focus in low light.
When we focus once or twice between 2 differently lit subjects, it all happens pretty quickly about a second or two, and we may not be able to easily measure the differences. When we focus between the 2 same subjects 30-50 times, it becomes much easier to measure differences in focusing speed.
I set up 2 targets, both were sets of Black Gaffers tape on a white wall.
Target 1 was positioned on a white wall next to an overhead lamp. In the manual mode, it exposed evenly at 1/60, 2.8 ISO 400 (this was confirmed with all 4 cameras). It was about 20 feet away. The reason I chose these settings was because 1/60 is the minimal shutter speed I like to use shooting portraits, and all three of my wedding lenses max out at 2.8. An ISO of 400 gives me some room to play. 
Target 2 was positioned on a white wall under a shelf, about 5 feet away. In the manual mode, it underexposed 3 stops for those same settings. (1/60, 2.8, ISO 400). From plain ol experience, I know most cameras should be able to expose properly on this target, even though it is significantly darker. 
All cameras were using Evaluative Metering, One Shot focusing mode (in the case of the D600 Single Servo). I turned on the “Beep” audio signal on all cameras to indicate focus lock, and I also turned off Nikon’s AF assist light for all tests (very important) – none of the Canon’s have an AF focus light, & I was interested in the native ability of the focusing systems in low light.
On the Canon’s I used the same Canon 24-70 2.8 L II lens. On the Nikon I used a Nikon Nikon 24-70 2.8 ED .
Aiming the test camera at Target 1, I would start a timer and immediately get a focus lock on a “+” gaffers tape mark, depressing the shutter button down halfway until I heard the audio beep confirming focus lock, & I would then immediately aim the camera at Target 2, again pushing the shutter button halfway down until the audio beep confirmed focus lock, then back to target one. For the main test on the center focus square I went through 50 focus locks, and here are the results:
Center Focus Square Speed through 50 focus locks:
1. Nikon D600 – 0:54 (54 seconds)
2. Canon 5Dii – 1:03
3. Canon 6D – 1:07
4. Canon 5Diii – 1:09
Between Targets 1 –> Target 2, the Nikon D600 was about 10-15% faster. The D600’s focusing systems are very good, but I was really surprised with the Canon 5DII . Its a very old camera, but that center focus square is really snappy and quick. The 6D and 5Diii feel like they are focusing in 2 stages, like a quick jump followed by a smaller, refined step. It’s probably safe to say that the Canons are all very close.I wanted to take this further, so I repeated the experiment, only this time, instead of using Target 2 I used a new Target “3” which was on the floor under my desk and required a 2″ exposure at f2.8 ISO 400, or ~7 Stops Underexposed with the same settings of 1/60, 2.8 ISO 400 It was also pretty close at about 4 feet, with Black Gaffers tape on a grey carpet. I know from 10 years of shooting experience that this area was dark enough that most cameras would have trouble focusing on anything here.
While I could still see the Target with my naked eye, it was so dark that if I shot it under the same test exposure settings, it was completely underexposed. Even processing in RAW I wasnt able to recover anything:
When I reshoot it with the 6D’s high ISO, we can easily see it.
I repeated the central focus point test from Target 1 to Target 3, this time with 30 focusing locks:
1. Canon 6D – 1:13
2. Canon 5Diii – 1:32
3. Canon 5Dii – 2:06
4. Nikon D600 – Failed to achieve focus lock after 6 attempts. (Note: D600 achieved focus lock in same tests @ 51s with its assist light, & the 6D also rated in at 52s with the infared AF assist beam from a Canon 600.)
I would say it is safe to conclude here that Canon 6D’s central focus point easily out performs all of the other cameras’ central focus points in low light, without asset beams
There is definitely something going on with the 6D’s focusing system on that central square. It is very, very good in low light, about twice as fast as the 5Dii & It even out performs the 5Diii’s central square.
The Nikon D600 failed to lock even once without its AF light.
UPDATE : Many readers were extremely bent out of shape over the fact I turned off the AF Assist Light, which may or may not be appropriate for the type of shooting you are doing. Similarly, there were questions about the AF Assist Beam from Canon Speedlites on these 3 Canon models, so I repeated the test. Very interesting results:
Canon 5DII (600 Speedlite AF Assist): 45s
Nikon D600 – Camera AF Assist on: 51s
Canon 6D (600 Speedlite AF Assist): 52s
Canon 5Diii (600 Speedlite AF Assist: 70s
Please keep in mind this is over 30 cycles, which translates to:
5Dii – 1.5 Seconds per Focus
Nikon D600 – 1.7 Seconds Per Focus
6D – 1.7 Seconds Per Focus
5Diii- 2.33 Seconds per focus
Whether this is something you can or cannot use is going to be up to you.
It seems to me, that when we combine the 6D’s very clean ISO noise, combined with it’s native low light focusing ability of the central square, it is an excellent low light performer. Think weddings & social events, street shooters will also love it.
Nikon D600’s focusing is going to be better for well lit situations, think any day time shooting, sports and wildlife. If you are ok leaving the AF focus assist light on, you will be in terrific shape.
It is very important to note that the 5Diii (41) & D600 (9) have many other cross type focus points which out perform the 6D’s other focus points, easily- so we can’t really say the 6D has “better focusing systems”, but we can say the 6D’s central focus point out performs the other’s central focusing points in low light.
I am anxious to read about any other independent tests others are doing and would be willing to repeat the test with constructive criticism.
Out of curiosity I also decided to test the very center top most focus square on Target 1 -> Target 2 test for 30 cycles -; I repeated for the furthest right center focus square:
1. Canon 5Dii – 0:36 / 0:43
2. Nikon D600 – 0:37 / 0:38
3. Canon 5Diii – 0:42 / 0:47
4. Canon 6D – 0:44 / 0:54
I also repeated the Target 1-> Target 3 test for the Top Center as well as far right center focus Square of each camera:
1. Canon 5Diii – Lock on 5th Attempt / Fail
2. Canon 5Dii – Fail/Fail
3. Canon 6D – Fail/ Fail
4. Nikon D600 Fail/ Fail
Well performed test. I highly appreciate it!
Thank you very much for your exceptional test! It would be very interesting to see whether the precision matches the locking performance on the 6D. Did you perhaps get any impressions on the matter? Did the test shots on -7 eV seem sharp on the 6D, at a first glance? Thanks again.
The "beep confirming focus lock" does not say anything about the focus accuracy acheived.
nice, thanks for taking the time to test.
Like Henrik, I’d like to see the precision of the 6D’s AF, but spefically the 6D’s outermost point vs. the 5DII’s outermost point. Using say the 50/1.2. But realise that’s a big ask! 🙂
Interesting test, but the results are quite expected given the AF specs of all of them. If anything it looks like Nikon understates a little bit the AF capabilities of the D600. For the sake of completeness, can you make a quick and dirty test of the D600 with the AF assist lamp turned on using the same methodology to have an idea of the absolute low light AF capability potential of all these bodies?
Thank you for the comparison. Some further test case proposals:
– Repeat some tests by using only horizontal / vertical tape
– Use rightmost crosstype point instead of rightmost point for 5D mark III. Distance from center is nearly same but performance could be much better in horizontal / vertical tape only test case
I found your article in canonrumors.com. It was an interesting test. However, you mentioned in canonrumors.com that your Nikon D600 test was based on disabling the AF illuminator but you did not specify here. I believe AF-assist illuminator exists in Nikon but not Canon, and most people with Nikon leave it on. Why would you disable this great function for your tests? The results were misleading.
The reason Michael didn’t use the Nikon’s focus assist light is probably because in event/low light work you can’t always use the focus assist lamp or the subject is too far away to be effective (e.g. night landscape).
Based on the fact that 0 EV is defined as correct exposure with 1 s, f.1 and ISO 100, the light level called "0 EV" in the test, according to my calculations, corresponds to 7 EV. Thus, the light level called "-3 EV" corresponds to 4 EV and "-7 EV" to 0 EV. This indicates that the 6D starts getting the advantage already on 0 EV and that it should be able to handle three full stops lower.
I myself rarely go below EV 3 or EV 2 so perhaps the 6D will not give me a properly noticeable advantage compared to my old 5D mkII. Of course also depending on the 6D:s actual accuracy.
Thanks again for your kind testing.
Surprised at the number of people who are choosing to ignore the heart of matter here over semantics, but do as you like- I love haters.
The hard part is getting all 4 of these cameras together, if you dont like what I did, please make a constructive criticism. Re-doing the test would be simple.
I turned the AF assist light off on the D600 simply because the other cameras dont have one.
Hi Henrick- How did you come to these calculations, they are a bit different than what I am hearing from other readers.
AC- The D600 actually does fine with the AF light turned on, but there are real -world cases where this isnt always ok to turn on. The other cameras dont have such a light.
biased test with D600 focus light disabled
Sorry, I didn’t realize linebreaks didn’t work.
I think it’s an interesting test and seems pretty fair. I don’t expect to shoot my D600 in pitch black. The dimmest I usually go is Target 2. If I’m shooting a show in some small venue with bad lighting, I still have adequate light to focus. AF-Assist light is a no, no in those situations, so people need to stop whining about it not being used during the test. Yes, Nikon does have that advantage in pitch black, but it depends where you’re shooting. Some places won’t let you use it. I commend Canon for giving the 6D this ability to lock in near darkness.
We have 2 5dii bodies and 1 7D bodies, and notice this same effect on the 7D. Using only the center focus point, the 7D seems more responsive than the 5D. How would the 7D do with your test?
Michael, would you mind listing the D600 performance with AF-Assist on and off respectively and let photographers decide how they should use it? I think a fair comparison is to use the best facilities a camera provides. Anyway, there will be no argument if all of the results are listed.
After the analysis of your review on CanonRumors, I’d have figured you would have updated this review by now. I hate to say it, but you are gravely misleading your readers as to the capabilities of the 6D. I really think you should re-do this test with a proper incident light meter, and base your EV0 on the official spec.
Hi Jon Rista- How would changing the measurement of light in these conditions change the performance of the cameras? What are you shooting with? Who is dying? (Isnt "gravely" a bit harsh?)
The engineers work for auto magazines must disable the anti-lock brakes of some vehicles to order to compare to others that don’t have it, and provide a "fair" review for the "native ability" of the brake systems. Very funny.
@ Brake Systems- You clearly haven’t read the test carefully. I have tested it both with and without the assist light, which the other cameras do not have. Depending on what type of photographer you are, this may not be an option. If a photographer cannot use it in many cases, why would that be an unfair test? Dont take it so personally.
Nice to see the different cameras against each other in low light
I shoot weddings with a 5D3; AF struggles in extreme low light conditions. Your test properly reflects that. Of course most cameras do, but the 5D3 and 6D can get decent shots (exposure wise) at ISO 25,600. IOW in extreme low light the 5D3 runs out of AF capability before running out of exposure capability. Speedlite AF assist doesn’t always help if you’re shooting a small subject. E.g, across a dark room at 200 f/2.8. The projected AF-assist grid pattern doesn’t zoom with the lens and is too coarse. In normal (even dim) interior lighting the 5D3 AF does fine, it’s only in extremely dim lighting this becomes a factor.
I wonder how the non-center focus points compare with the D600 for low light. I am curious because I am really tempted to move from my Nikon D80 to the canon 6d for the extra 2 stops ISO over the D600 but I wonder at what light level the non-center ones start to fail.
Ive done tests with the non-center focus squares, and can tell you that they are no good in low light. It is only the center square.