A variety of methods, all of them of the iterative bootstrap ilk. Star clusters can be used, so if you find variables in a cluster, you can get the distance to the cluster, therefore other non-variables in the cluster as well. There are period luminosity relationships established for other variable types besides Cepheids, even novae. With this method, you can correlate spectrum details with intrinsic brightness, which can then be applied to similar stars not in clusters.

Another cluster-related method uses nearby clusters such as the Hyades, whose 3D motions can be reconstructed from their proper motions and radial velocities, and the vanishing point effect caused by their traveling together (think of a cloud of buckshot whizzing by you). The paths of the stars are approximately parallel, but the perspective makes them appear to converge, so geometrical analysis gives you a distance. With a distance and an apparent brightness you can derive intrinsic brightness and then correlate that with the spectra so stars of similar spectra elsewhere can be assigned a distance.

In general, as we learn more about stellar evolution, we can correlate spectra with intrinsic brightness, which gives us distances. A G2 giant will have a different spectrum than a G2 dwarf, even though their surface temperatures and spectral classes will be the same. These spectroscopic parallaxes can also be supplemented by what are called secular parallaxes, which use the baseline of a piece of the sun's orbit about the galaxy instead of the earth's orbit around the sun. The astronomical jargon for distance is "parallax", whether it is derived by triangulation or not.

Needless to say, all these methods interact with each other and influence one another, so an improvement in any one technique alters our values for the others. For example, more accurate studies using the Hyades "flying cluster method" give better brightness/spectra data which in turn alters are ideas about stellar evolution which in turn suggests refinements to our standard candles. Every now and then, some new technique or improved observation gives us better data and corrective changes propagate throughout the entire system.

If it all sounds like a house of cards, it is, and astronomers are painfully aware of it. But over the years results from independent methods have been converging and the "adjustments" are getting smaller, so they have confidence in their distances, even if they concede they are not very precise.