I'm posting this information in case you may run across a similar problem.

PROBLEM
1998 Cadillac DeVille with Northstar control system and with
intermittent MIL, sometimes flashing.  Vehicle was taken to dealership
and P0303 code extracted (cylinder 3 misfire).  Number 3 fuel injector
replaced and all is quiet.  MIL turns on, again, sometimes flashing.
Vehicle taken to an independent shop, P0308 (cylinder 8 misfire) code.
"Motorvac" treatment given, problem goes away.  MIL turns on, sometimes
flashing.  Vehicle taken to dealer, DTC P0308 present.  On way to
dealer, MIL turns off.  Customer states the problem often goes away.
Tech performs power balance test and recommends replacing #8 injector.
Customer  refuses, claims dealer  incompetent.  Customer pays $150
diagnostic fee and leaves.  Since "customer" is a family member, I get
to look at the car, next.

EXPERIMENT
Scan tool (OTC Enhanced Monitor with Pathfinder 99 firmware) showed
P0308 in history, both long-term fuel trims were negative (as expected
on a hot engine -- although one was more negative than other).  Vehicle
appeared to be running well when examined.  Ignition on cylinder 8
'scoped and compared to "known-good" (I used #6 since this was
convenient and PCM was not complaining about this cylinder).  Both
cylinders were firing at about 10 kV.  "Good" is between 10 kV and 15
kV.  Decorative top cover was removed , tape from #8 and #6 fuel
injectors' wiring was unraveled and current peak and duration measured.
Both injectors were drawing 1 ampere, pulse width was about 3.2 mS
(scan tool showed 3.3 mS was commanded).

Pathfinder 99 allows individual injectors to be turned off.  With engine
off, idle air control motor was disabled and engine was started.
Selectively turning off injectors showed approximately identical RPM
drop (before fuel control kicked in).  Engine was turned off.  Idle air
controller was reconnected and new DTC ("cannot control idle rpm") was
cleared.  Although the power balance test was not done "by the book,"
its outcome convinced me that, at the moment of the test, no problems
existed in #8 cylinder.

HYPOTHESIS
System appeared to be in very good control of fuel delivery, although
slightly different long term fuel trim (LTFT) readings hint at a
possible air leak (injector O-ring, intake manifold, etc.).  Misfire was
not present.  Initermittent problem is symptomatic of electrical
connectivity issue, characteristic of fretting corrosion (due to
microvibration inside electrical connector and oxygen ingress causing
oxide formation -- do a web search if you're interested in details).

REPAIR
#8 fuel injector connector was disconnected and injector's pins and
connector clips cleaned with Caig Labs' "Deoxit D5" (a very good contact
cleaner/lubricant) and reconnected.  Connection reopened, recleaned and
final connection made.

FOLLOW-UP
Vehicle has been running flawlessly for three weeks.  Family member told
next time P03xx code present, all injectors will have connections
cleaned.  Total time spent on job:  2 hours.

If you want to contact me on details, please replace the digits with "o."

Cordially,

Bohdan Bodnar