It is indeed mismatched, in the strictest sense of the word, which implies optimum power transfer, which occurs with a loss of 50%. Microphones are supposed to be bridged, i.e. optimized for voltage transfer. There is no strict definition for bridging, although it is commonly admitted that the load impedance should be about 10x the source impedance, resulting in an input impedance of 1.5-2k for a mic. 600r reflected impedance clearly does not satisfy the predicament, which doesn't mean the performance will be necessarily degraded significantly. Mic designers optimize their products in view of the 10x rule, however, there's a recent tendancy to increase the input impedance of mic inputs, as in some cases they would measure at 10-20k. The audible benefits are not always obvious nor decisive. However, IMO, reducing the input impedance never gives interesting results, sonically.
I should mention that in most cases, variable impedance is achieved by shunting the signal with a potentiometer, which is an objective degradation of performance. Academically, variable impedance should be accompanied with simultaneous adjustment of the operational parameters of the circuit. One can consider the Neve 300/1200r switching as a better alternative, since it changes the input xfmr ratio at the same time, but it doesn't change the input transistor's quiescent in order to optimize the OSI (Optimum Source Impedance); the tolerance to source Z is vast enough to make that unnecessary in practice, though.
Tube mic pres present a specific case, as the secondary is supposed to be loaded only with the grid of the input tube. As a result, the impedance presented to the mic is variable, since it depends very much on the raising reactance of the primary's inductance, resulting generally in a gentle attenuation of LFand upper HF resonance, until it hits the limit due to the Miller capacitance of the input tube, which can be as high as 100pF, as PRR mentioned, for a typical triode, or much lower, for a pentode, or a cascode stage.
The gentle brightening of the frequency response is a fundamental design point in RCA ribbon microphones, where the load is specified as transformer with unloaded secondary connected to the grid.