The microwave oven transformer is one of the best power supplies (in my opinion), it is cheap if not free, can deliver a lot of power and relatively small unlike large potential transformers or pig pole transformers.
A single transformer delivers about 100mA at about 2KV, this is a reasonable low voltage and you will probably need to do something with it before it can be used.
There are 2 common solutions to this problem, the first is to connect multiple transformers together, this is a simple but effective solution as the current output remains the same, the alternative is to construct a voltage doubler or tippler. This will give you a higher voltage but the current output will be reduced and their will be losses in the additional components.

The biggest problems with using a MOT is current limiting, unlike NST's which can run shorted for ever and ever a shorted MOT wont last very long. MOT's are engineered to work with a magnetron nothing else, when connected to a magnetron they will work perfectly drawing the correct current, when you try and use them with something else you will have to find a way of limiting current flow.
There are various solutions to this problem many of these involve quick solutions such as connecting a mains light bulb in series or an electric heater, salt water resistors are also used. I usually like to go for the simplest solution but I also like a decent setup which rules out the for mentioned options.
The remaining solution is to use a ballast or an inductor, this is a coil of wire that limits current flow; as the current changes it sets up a magnetic field, in turn this magnetic field opposes the flow of current limiting the current at some value determined by the inductance of the coil.
The simplest inductor is another microwave oven transformer with the secondary shorted out, the primaries of the 2 transformers are then connected in series. A more professional option is to wind your own (or have custom built) inductor; simply take a transformer core (an old MOT) and wind more wire around it.

My MOT power supply was made from 5 MOT's, 4 connected in series and 1 connected as a ballast. The 4 MOT's primaries were connected in parallel, this gives full mains voltage to each transformer, the secondary's of these transformers were then connected in series, this adds up the voltage of each of the transformers, the final MOT was then connected in series with the mains (the secondary was shorted) this provided relatively good current limiting.
During testing I found that the supply drew up to a maximum of about 14A at 260V (the output of a variac), that's 3.6KVA; the supply outputted about 8KVAC and about 11KV DC with a current of about 0.5A, this is a lot of power. The arcs are very impressive burning with a very hot flame.

Blasting a MOT
Below I have detailed how I went about trying to design an inductor to limit the current in a MOT, in the end I just used a shorted MOT.
07/02/05
I ordered 2 MOT's from a surplus sales company for £8 each, I am sure there are many of you who would call me an idiot for buying something that you can get for free; yes you can get them for free but I simply don't have the time to find old microwaves. The plan is to test out the above approach of connecting the 2 in series and shorting one of the primaries, I will have some pictures of this when I get around to doing it. I then hope to carefully dismantle one of the transformers and wind my own inductor, something decent and reliable that I can use for different applications.
Dismantling the transformer unfortunately wont be that easy, they are constructed from E's and I's, imagine putting the I next to the E and you get a perfect transformer core. The E's and I's are very thin laminations that are stacked together, once they have been stacked they are welded along the seems. In order to reuse the E's and I's I will need to carefully cut through the welds with a junior hacksaw, making sure I don't damage them that much. Once they have been removed a frame will have to be built to hold them together and vary the spacing between the E's and I's.

I have simulated a possible inductor design using an electromagnetic design program called MagNet, I don't know the dimensions of the transformer at the moment so I built a 10x10x10cm core and wrapped a coil around the centre coulomb, I then tested how different numbers of turns affected the current and the spacing of the top I from the main core
The current varies hugely from 500A with 10 turns down to 200mA with 500 turns, I decided on 500 turns as it stated me off with a relatively low current and allowed me to increase by varying the position of the top I.

As you can see from the image the flux in the centre is very high reaching 3.8T this saturation is preventing the current from flowing

I then investigated removing the top piece to see how this affected the flux, with a 1mm gap the current jumped up to 1.81A, and then increasing to 6.4A with a 10mm gap

In this picture you can see a 5mm gap, with the same 240V flowing through the 500 turn coil, the flux had dropped to 0.84T, significantly lower than with no gap, you can see how the flux "sprays" out of the gap in the magnetic circuit.

These results were only a quick simulation with an estimate on the core size, when I have taken apart one of the microwave transformers I can put some exact measurements in and see how it really behaves.

11/02/05
I found out that my 2 transformers wont be arriving until Monday, I was hoping to test them this weekend so I have decided to take apart the old MOT I already have, I am not going to connect it up I just plan on dismantling it and trying to do something with the core, then on Monday I can put the exact measurements into MagNet and design my inductor.
This is the MOT with the windings still on the core, you can see the seam on the left where the E and the I are joined and the weld holding them together

15/02/05
With the help of one of my lectures I have modelled the inductor using the exact core dimensions and highlighted a big problem, the core I was planning on using is too small. I was planning of varying the current by varying the E-I spacing, unfortunately for a current variation of 0 to approximately 15A I would need 500 turns of wire, this would then need to be thick enough to carry up to 15A, given the size of my core this impossible, it would just about take 500 turns of enamelled copper wire rated at just less than 1A. I have now decided to use this as a fixed inductance core for limiting MOT's at 2A, this would require only 50-60 turn of wire.
I will then hopefully get a bigger core so I can design my large variable inductance inductor again.

24/02/05
Unfortunately the 50-60 turns doesn't take into account the saturation of the core, I didn't think this would be a problem but unfortunately it is, I have now come up with a new design, it requires 350 turns and a 0.5mm E-I spacing. The E-I spacing is designed to reduce the core saturation and the windings have been increased to compensate for the reduced inductance. This should reduce the current to 2 Amps. I have bought some thin enamelled copper wire which should be suitable, I have also ordered an 8A Variac so I will be able to test these designs beyond 2A, hopefully I will try this at the weekend.

03/03/05
I wound the inductor using 1mm enamelled wire unfortunately I only managed to get 225 turns on the core. I remodelled this design and found that I should be able to limit the current to about 5A, unfortunately the wire is only rated at 2A. I will try increasing the current bond 2A and seeing how it heats up, for the testing i have been connecting the inductor straight to the variac, this is giving me a pure inductive load causing little or no heating, in order to correct this for more realistic testing I will connect a capacitor (probably a microwave capacitor) across the inductor, this should help bring the current in-line with the phase.