How could 100 Watt class a has more head room than a 300 Watt amp Class AB

Eh... The drawback of SE class A is you don't enjoy the benefits of distortion cancellation without feedback or lesser feedback unless you use truly exemplary gain devices. The class A envelope also ends up being the hard limit of the amp regardless of power supply because the topology transitions into class B operation instead of AB. That's why I assumed the OP was referring to push-pull topologies. 

ML is in the past, Gryphon is in the present. Bloody expensive stuff, though.

A Class A single ended amp gets rid of whatever crossover distortion might exist in push pull circuits so there’s that, and when exposed to this phenomena the listener is rendered a better person almost instantly.

+1 for that, and that’s any class-a not just SE.
Ever listen to a pair of old ML-2’s in good working condition through good ESL’s with maybe dynamic bass like Monoliths, MAJESTIC!!!!!! So long as you don’t want party levels.
This amp keeps doubling down to 1ohm, I’ve seen it do it on the bench.
Cheers George

 The sound of music (wasn't that a movie?) when it reaches your head is far more important than the wattage your amp produces, so if it sounds good at 12 or 2000 watts, it's good.  A Class A single ended amp gets rid of whatever crossover distortion might exist in push pull circuits so there's that, and when exposed to this phenomena the listener is rendered a better person almost instantly. 

'Headroom' can have a lot of meanings in the audio world...

At system-level you may think extra db from the average listening level the system can play before non-linearity/distortion, for example. This does involve the amplifier, but all other elements also including acoustics.

Headroom for amplifiers can mean extra power capability above rated, typically for shorter peak periods. But that's not all the possibilities for the meaning.

If you are using speakers that reach down to low impedance at some frequencies (as mentioned by georgehifi previously), power supply / output stage current capability and output impedance may be way more important that rated output power. First Watt Class A is a great example of that type of amp that can drive low impedance easily w/o fatigue on difficult loads.

But it all depends if that's a factor in your particular speakers/system/listening style. Really. An apple may not be the right fit for a person who is looking for an orange...

Unfortunately there's no general rule of thumb figure, it's all situation-dependent.

dont know excatly what type/topology of amp you're talking about. I am sure that it would not be the case with some other brands/models. But it would generally be explained very simply by the fact that the AB amp has less power reserve, or the constructor is just fooling customers, or something else in the specs has not be taken into consideration; output impedance maybe...

my "shoot from the hip" answer is that the one with the more robust power supply will have the most headroom and be capable of the most dynamic swings

you 'presumed' but apparently did not actually 'measure', a classic audiophile mistake. You said the 100 watt amp 'appears'.........means nothing in this scenario.

To compare meaningfully output power of two amplifiers (all external factors being equal) it is useful to know several parameters.
--the manufacturer's definition of the power output metric. It should be
   "continuous power", meaning the amplifer's delivery  is not time limited  and it does not overheat in the ambient temperature 
--what is the distortion 20hz-20khz
--what is the signal/noise ratio
--what peak output power can be reached at 2-4-8 ohm, for how long, at what frequency and with what distortion
The peak parameters are useful to know, as musical peaks may need 5-7 times value of the continuous output power.
  

You've got a point, if we're here talking apples to apples, but where not. We're talking totally different classes of amplification. My 41 watts class A amp will transition to class AB and push WAY more power than 41 watts. It can reliably bang out about 150 watt peaks for brief before reaching the limits of what the transistors will tolerate, and that's only really limited by the sinks accumulating too much heat. Yeah, the distortion does get a little high, but not horribly so. A 100 watt class A push-pull should be able to effortlessly drive about 300 watts into a load without burping. 
But think about that kind of power for a minute. 100 watt peaks are MASSIVE on half way efficient speakers. Unless you're driving horribly inefficient speakers in a mammoth space, are you using anything like 300 watts? My 41 watt F5 practically never leaves class A regardless of how much Bassnectar I throw at it. It idles at 54C no matter what I throw at it. 

Gentlemen,

Unless you are using a sound level meter to determine the actual sound pressure level what you are probably experiencing is the following phenomena which is that we think something is playing louder when it starts to distort and don't think it is too loud when it is not distorted.

Under circumstances of no instrumentation to aid your perception the 100 watt amp sounds louder because it is distorting sooner, the 300 watt amp sounds less loud because it can play louder with out distortion.

Your brain and ears are deceiving you..

Mastersound PF 100 Monos. 110 class A SET. 
Quite an experience. Take a control over the speakers unseen for me before.  The only amplification I can confirm that kef reference are musical. 
Simply amazing 

I've got sitting on my table a clone of a First Watt F5 running hot rails of 32.5V and biased up to 1.26 amp. In push-pull, that's 2.52 amp by 32.5V for a class A envelope of about 82 watts. 41 watts RMS. Keep in mind that when you halve the load impedance the amp will double it's output, but it will also halve the class A envelope and push it into class AB operation sooner. 
The power supply really is the crux of the matter. A muscular class AB amp might only call for 50,000uF of filtering, and may only require a 600VA of transformer. I think that's what the Schiit Vidar is packing. I've seen guys building F5T amps with dual 800VA transformers and and 480,000uF of filter capacitance. My F5 has a 600VA toroidal and 120,000uF of filtering for 41 watts output stereo, and there are guys that build that amp with one of those supplies for each channel. Class A amps require absolutely massive power supplies because they're always pushing the full class A output power through the circuit at all times. You'll never find class AB amps with power supplies like class A amps. The class A envelope on mine is only 82 watts, but it will transition to class AB and handily push over 300 watts RMS until something melts. As it is, it barely ever leaves class A driving my Focals. 100 watts in a well built push-pull class A amp is an absolutely massive amount of power with huge headroom, likely far more than any class AB amp of triple the rating. 

Take this into account, at normal moderate listen levels, a pair of Mark Levinson ML2 (BJT) monoblocks are only 25w @ 8ohms, but they will out perform things like a Perreaux 3150 300w @ 8ohm (Mosfet) into hard to drive speakers, because the ML can deliver more current into that harder load, where the Perraux will start to compress.

Cheers George

We also need to take into account the output impedance at the rated power. depending on the speaker load, a 100 W amp might play louder if it has a lower output impedance than the 300 W amp, especially if powering a nominally low impedance speaker. 

@elevick 
You are correct,  I just threw out 6db because of going from 100 to 300, but yes, you would need 400 watts for the EXTRA  6db of headroom.
I do disagree to some degree on not telling the difference.  Not necessarily but could be.... I have heard a 40 watt amp that had a huge power supply that was so dynamic that if you had fairly sensitive speakers that you would easily believe that the amp was a lot of power... On the other hand,  regardless of power supply.  If you had a very low sensitivity speaker,  say 82 to 84 db and you liked listening at a decent level, ALLTHINGS BEING EQUAL  you wouldn't be hard pressed at all telling the difference between 100 watts and 300 watts. Of course, you didn't say all things being equal and I do agree that overall, I would prefer a 100 watt Pure Class A amp to a 300 watt mediocre Class A/B

Timlub, sorry to do the math for you but 100 watts will gain 3db every time the wattage doubles.  200 watts +3db, 400 watts +6db.

You will be hard pressed to tell the difference in amp power when comparing most a really good 100 watt class A to an average 300 watt A/B.  Check the specs also for dynamic headroom.  That might be your answer.

Not having any credentials such as a manufacturer or scientist, I always refer to the basics.For me it adds even more confusion, because we're talking about a music signal heard ultimately, by ears.

http://www.electronics-tutorials.ws/amplifier/amplifier-classes.html

I took a couple of analog/digital classes in the early 80s. An instructor that was an audiophile would blow the students minds  with endless physics/science  theory behind audio. It was way over my head!

I did become aware that most of us really are listening to just a couple watts of "linear "power. 

There are a few examples...So as far as Headroom.... First an 100 watt amp that is true pure Class A amplifier won't play louder than a 300 watt A/B amp... Pure Class A means that it will not switch out of Class A from First Watt until clipping.... So the amp is a Class A amplifier... In this sense,  A  100 watt vs a 300 watt amp is a no brainer as far as how loud it will play...The 300 watt will have 6db more headroom. On the other hand,  there are several amps today that are sliding Class A,  They are biased to run Class A for a percentage of their output.  Threshold and Coda did this quite a bit.... I think of the Aleph 3 or 30,  they are rated at 30 watts class a, but in reality are more like 125 watts or so,  but they play their first 30 watts in Class A before switching to A/B, after 30 watts they play the balance of their power in A/B. 
I hope this helps,  Tim

Sounds like you are talking about two SS amps, so probably from Pass but that is just a guess.
To your question, the Class A amp will likely have a stiffer power supply and more capacitance.  Also, IME, I like the sound of Class A better since to me they sound more fleshed out than the best AB amps I have heard and so may sound more harmonically complete when pushed.  A last observation based on experience is that really good Class A amps do not get strident sounding as Class AB amps can when they are starting to get stressed, the Class A amps simply run out of steam - at least that is how the two different sets of Lamm hybrid monos I owned used to behave.  My 300 wpc Class A Claytons have yet to run out of steam, even driving my new lower-efficiency (85dB) speakers.
To George's response, my listening has resulted in a preference for bi-polar output stages vs. mosfets.  I believe some designers use mosfets to try and emulate a "tube sound" but to me they do not control the output as well.

How could 100 Watt class a has more head room than a 300 Watt amp Class AB

If one were a Mosfet output and the other a BJT (bi-polar) and both were well engineered push/pull and driving into a heavy load, such as the Wilson Alexia which has an EPDR (equivalent peak dissipation resistance) of just .9ohm!!!! at 65hz.
Then the 100w BJT amp would drive it better than the 300w Mosfet.
As the 100w BJT "could" in effect give out 800w into that .9ohm load where the Mosfet would probably go down to less than it’s rated 8ohm 300w output.
Cheers George

Single ended class A amps don't so I'm going to assume you're talking push-pull class A. A class AB amp is usually said to clip when the demands of the waveform exceed the voltage and/or current capability of the amp circuit or power supply. Usually the power supply is the major limiting factor. 
A P-P class A amp is roughly like a class AB amp, but the bias is set WAY higher. Instead of a few milliamps just to keep the devices on in their linear region, the bias is cranked up to cover the entire waveform. Once a 100 watt P-P class A amp reaches it's bias limit, one side of the push-pull goes completely off and the other side conducts beyond it's bias. That's a p-p class A transitioning into class AB. That's actually how high bias class AB amps are designed to work. The result of the p-p class A amp behaving like that is massive amounts of head room. As Erik suggests, loads and designs strongly contribute to this, but that's the general reason why. 

You are going to have to offer very specific examples of the amps you are talking about as well as the speakers.

Best,

E