Specifications and reliability. Industrial vs. commercial spec parts, wire wound resistors, temperature compensation devices, 5ppm oscillators, lots of reasons.
Some of the old tektronix wiring really is a thing of beauty. The tektronix 360 modules are so lovely.
A few years back I managed to buy a cheap digital voltmeter that was actually wrong. As in: consistently off by 20% or more when measuring voltages in the range of volts. (!)
Ever since I've been trying to figure out how much to spend to get a voltmeter that is guaranteed not to suck. I figure it probably still needn't be very much, but I still haven't chosen one, because I've managed to spook myself. I tell you, on the day after you catch your test equipment giving you egregiously bogus misreadings it is hard to resist the impulse to spend the money on a Fluke or Textronix.
I bought a $50 microcontroller based DMM (much better than "dumb" dmms) from circuit specialists forever ago here:
http://circuitspecialists.com/prod.itml/icOid/9455
It works really good. Comparable to fluke handhelds but not in the same league as the fluke 4/5 digit bench DMMs.
Keithley stuff is great too, more science based though. I've used everything from their picoammeters to their general purpose DMMs.
Apart from the accuracy, build quality, etc. one of the reasons I bought my Fluke DMM is that I was working with some high power circuits (like 600VDC at 60+ amps) and I trust the Fluke more than a random cheap meter not to blow up. There's a reason the fuses for a good multimeter cost more than the average discount instrument.
Some of the old tektronix wiring really is a thing of beauty. The tektronix 360 modules are so lovely.