Well, Anthony B is probably correct that magnetometers are not much use at finding caves in Limestone, but you can get caves (quite big ones) in other rocks, and at least in the case of lava tubes magnetometers appear to be exteremly useful in detecting voids. Although not conclusive from the point of view of anyone actually entering the caves as yet, Caesium magnetometry has been used in Iceland to predict the existence of at least 400m of passage beyond the lava seal that ends the large Stefanshellir system. The results are definitiely not ambiguous, and the Icelandic Speleological Society have even been so bold to name the cave - Hulduhellir.
Though I was involved in the field work, I am not expert enough to know how this works, other than a few sketchy details. Essentially the magnetometer picks up the interface between the rock (which became magnetised when it was liquid) and the air in the cave. This provides a specific response typical of cave passages. It shoudl also be noted that the Hallmundahraun lava flow provides almost ideal conditions for employing the technique with little surface relief and large (20m diameter) tubes only 10 or so meters down.
If you want to learn more try tthe two Laki Underground reports from 2000 & 2001 authored by Dr Chris Wood et al from Bournemouth University / Shepton Mallet Caving Club.