Hmmm. Very carefully laid out grounds all tied back to a buss made
from 12 ga. household solid core wire. Logical progression of stages,
without criss-crossing of leads if at all possible. Also, lots of LOCAL
power supply decoupling caps. Get 'em as close to the stage they
serve as possible, which will localize signal currents on the power
supply rails.

There are only two places where there is intentional HF shunting to
ground in order to prevent oscillations. And there is plenty of HF
going on too.

Keep the nodes which have mega signal voltages on them
SMALL--for example, if you're going to use a votlage divider to
knock some AC volts down, mount the series resistor as CLOSE to
the coupling cap as possible. This way the node with the highest
voltage swings is contained in a small area.

Don't be afraid of the 3-D layout--use all of the space in your amp,
and be sure to float those coupling caps in the air. They give off quite
a field.

For some idea as to how my preamp looks, go here:
http://www.geocities.com/SunsetStrip/Garage/5701/preampmods.html

This was by far the most successful layout scheme I've tried yet.
And the layout is probably the most critical thing of all.

1) keep input wires as short as practical
2) keep output wires away from input wires
3) keep input wires away from oscillators, 
and components driven by gain stages.

In the case of a tube-driven effect, use DC heater current or twist
and isolate the heater wires from EVERYTHING else in the box.

The goal is to keep the wires from acting as antennae. 

 Resistors
 always try to use metal films. What is the HIFI argument against
them? Carbons are just too noisy (IMO). That was the entire
noise problem in my Traynor; It had carbons. I switched them out
to metal films and no more noise with no negative effect on tone
whatsoever, as a matter of fact, the overall response of the amp
got sweeter, not as a noticeable listening effect, but as a function
of feel and interaction with the amp.