My own approach is a bit complicated and is outlined
here. Of course, most of the complication is because I want battles to work out more like those of WWII.
Ignoring the complicated interdiction bit, I paralleled the real-world comparison of aircraft vs watercraft (namely, the latter being far more efficient) by having two different types of drives - fully reactionless and semi-reactionless (basically, reaction drives with impossible efficiencies). The fully reactionless versions have such atrocious power requirements that you cannot use a proper power plant for them - rather, you need some high-output, low-efficiency power supply. The semi-reactionless ones require very little energy and reaction mass, but give much worse performance.
The result is that vessels with reactionless drives are incredibly maneuverable and have excellent acceleration/velocity, but are rather short-ranged. Vessels using semi-reactionless drives are more ponderous, but they have incredible range. This gives a good split between fighters (small craft using reactionless drives) and capital ships (larger craft using semi-reactionless drives).
This scheme
does allow for a massive SM+15 behemoth that nonetheless handles similarly to a much smaller fighter. It would probably be limited to planetary defense, or need to be hauled to the battlefield. If this is problematic, you could go with weby's suggestion of the high-acceleration drives having a maximum size limitation. One good justification for that is stress - a higher mass vessel is going to suffer far greater stresses during high-G maneuvers, so you might have a size-based limit on how quickly a vessel of a given size can go before it risks tearing itself apart.