• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

half -ife

Power tubes are by far the most likely to blow. Small signal tubes are (properly used) good for nearly forever.
Any part, whether electronic (tube or solid state), mechanical, or what have you, will show a similar pattern of failures. If you chart them on a graph (number of failures on the vertical axis vs. time on the horizontal axis) you end up with the infamous "bathtub" plot, thusly: \___/
In the beginning, you're going to have the 'infant mortality' cases. This will include devices where something went wrong in the manufacturing phase or gross mishandling during shipment. You then enter a long, stable period where the parts perform well and failures are few. Finally, age takes its toll and parts begin to fail purely because of the cumulative effects of time and use. Electrolytic capacitors are a good example of this, as even a never-used cap will pop the first time voltage is applied, if it is old enough. Note that the relative proportions of the beginning, middle, and end of the curve change from device to device; some are more robust than others, some are more prone to age effects, etc.
Yes, the circuit itself can be rough on a part. There are a lot of people on this site who've built one or another of Nelson Pass's designs. Since many of Nelson's designs are class A, the parts are run hard. This explains the frequent threads from anxious builders who are worried that part X may not be robust enough to take what the circuit is dishing out. Heat also factors into the equation, as heat dissipation is a major part-killer. Tubes, bless 'em, aren't all that worried about heat; give them an inch or two on each side and they'll radiate their own heat. On the other hand, the circuit design can be just as rough on tubes. As an example, when I built my tube amps I first ran the 6550's at 55 mA bias current (575V rail). The 6550 is rated for something on the order of 34W plate dissipation (I'm quoting from memory, here--don't shoot me if I'm off a bit). Since I was running somewhat less than 32W, I thought I'd be safe. Not so. My bias was regulated, but my rail was not. My AC at the house varies enough (curses!) that it would run the rail hotter than I had designed for. I popped tubes as a consequence. I now run at 50 mA each and haven't lost a power tube since.
One of these days, I'm going to regulate the rail, but that's going to be a nuisance due to the high voltage (the bias [as noted] is already regulated, as are the separate rails for the first and second stages; the main rail is the only unregulated DC in the circuit).

Grey