What are you probing?
I'm building a tube radio right now. I've been doing stage-by-stage tests by terminating their outputs into 50 ohms directly. This has two advantages: the dynamic range is more than enough to get a useful signal out (typical tube currents are around 10mA, giving on the order of 4dBm; my HP 8590A has about -70dBm noise floor at typical settings), and I don't have to worry about the output parameters of the stage (compared to the resistance of kohms+ and reactance of ~300 ohms, typical of the output terminal at this frequency).
If you're testing tuning/matching networks, it can be very useful to build them to match to 50 ohms. For example, you might have an impedance of ~1kohm from a transistor collector/drain, which obviously cannot be observed with a 10x resistor probe, nor is it practical to use a scope probe (with ~10pF of capacitance that will completely mess up your frequency response) -- even if you had the 1M to 50 ohm amplifier box to use such a probe on the spec.
If you're matching one stage into another, with absolutely no need to touch 50 ohms inbetween (it would be a waste of materials and time, and you'd have to re-tune it anyway to optimize out the 50 ohm link), it might be better to wire up the following stage as normal, except leaving its output shorted into 50 ohms. What better probe than the device itself? And you get gain for free, besides!
In general, transistors are quite happy being shorted into a low impedance. The output characteristic is largely capacitive, so you only need to check that the baseband bandwidth is sufficient (Fo = 1 / (2*pi*(50 ohms)*Cout) for the frequency you're monitoring.
That said, because transistors operate with voltages much lower, and have much higher gain, than for my case with vacuum tubes, you won't be able to assume so easily that the output is acting like a true short. Miller effect and all that will still be relevant. But with it being so easy to build grounded-base/gate or cascode stages, you can use those to your advantage, for this reason and others.
Tim