Copy pasted from an old e-mail, TL;DR, HTH and all that:
T5s input shaft/5th gear swap with garden variety gear pullers/various orielly/harbor fright tools is straight forward enough & well documented with those having infinite aftermarket, internet documentation & support.
T5 turd polishing from the myraid of T5 applications can rapidly become cost-ineffective, however.
Challenging to shop / choose wisely amongst often beat up used transmissions driven in a (likely) thrashed mullet missile (Camaro/Firebird/Mustang, typically). All other known applications in the USA have ratios about as terrible as an M46/47 / are about as weak and often a lot more beat/used up with more power and/or heavier car and/or abusive/neglegent previous owner than even the OE volvo trans; cheap, but not necessarily inexpensive. Eclipsing new T5 price, good used condition WC 2.95-.72 price with (more) persistent search(ing) and/or some other stronger trans all together is easy enough while trying to improve undesirable T5s, ending with poor/costly (TB budget sliding scale) results is easy enough, however. BTDT once, but not twice, mercifully.
North America (Canada?) never received the WC 2.95-.80 Cosworth imported en masse, but the parts are available new & used nonetheless, of the inexpensive choices.
The "next closest OE ratio spread" (2.95-.72) WC T5 (USA market) was only offered in some IROC V8 Camaro/Firebird 1988-1992. Some factory production volume, but the vast majority were:
(new production stats/drivetrain offerings)
-Non-World-Class before 1988ish (very few parts compatible between NWC & WC, but NWC and WC isn't necessarily of great significance for raw torque handling/abuse tolerance capability by itself, but for any upgrades it's of significance & greatly limiting/rapidly cost-ineffective...you're pretty trapped with an inferior parts bin to recombine with if you get a 1983-1987 NWC even in 2.95-.72, but acquisition price is accordingly (usually fairly pathetically) lower (these days).
-V6/terrible ratios
-Automatic
.63 OD/5th w/V8 to meet avg. C.A.F.E fuel econ NOX regulations is standard in most, even if you're lucky enough to get the V8 2.95 1st. Good/great for fuel econ/emissions in the light-weight fairly aerodynamic Camaro application with all torque and no power at 1500RPM, but basically worthless for anything but cruising in the flat at 80-85+mph in ideal conditions in most other applications.
-Only offered in the most expensive (that avg. joe didn't/couldn't buy) &/or most customized specially ordered models / rarely sold as floor models (most of those were stripper pathetic V6 loss-leader fodder and/or loaded with an auto-box/all the toys).
(25-30+ years later...):
-Often driven hard/long since wrecked
-Very nearly every S10 V8 conversion guy and/or other Ford / Chevy guy goes to the junkyard/auction at 6am/crack of dawn all the time to pull them/ canabilize for parts or complete. Competing with those guys isn't easy as many will take/use just about anything viable/usable whether they're replacing what they broke or upgrading.
-Engine torque near/at torque rating of trans when new.
Examining production numbers, and/or drawing a matrix of just those basic new drivetrain options & calculating the baseline probability/knowing nothing else, lottery odds already aren't great. Add more factors / 25-30 years it's difficult to interpret the odds as 'improving.'
Using later (1994+ body style) V6 Mustang donors and remaking 2.95-.72 ratio spread with some/all new used parts is often effective, at this point.
With all the costs of transmission/drivetrain adaptation to the redblock 4cylinder (T5 or otherwise) (~$1k+ observed repeatedly...no doubt a heroic story is brewing / can be found with a captivating plot outline w/ analogous storyline "something something something barn find adapter plate for .02?/ stored in a hyperbaric chamber in NOS condition, suitable clutch that works perfect for all uses and/or all the (minor enough/kludgery) fab work done for free/.17?/hr"), why settle for lousy ratios/overall strength?
Band-aiding a mis-match rear axle ratio in that's special tool/labor intensive/technical if setting pinion depth (likely the average costs-conscious TBer obtains a common 3.73 rear axle complete/"known good" many times, in fairness ) with the 3.3 1st T5 seems completely backwards.
1031/1041 R&P aren't likely to have any strength issues (even with slightly smaller 4.10 pinions, the axles &/or diff carrier always breaks first it seems/that isn't nearly as much of a problem piece of strength as the transmission/other differential parts. 2.95 T5 parts bin "best of OE" is less likely to break behind a 4-cylinder with 4.10 gears than weaker 3.31 T5 with 3.73s that's being crashed around/beat on with greater rev-drop between gear changes to be competitive and/or 'fun' in most "real world" conditions.
3.3 first w/.68 5th is FoMoCo frugal T5 parts bin recombining so they could (logically/understandably) meet (incoming) C.A.F.E fuel economy & NOX as well as obscure granny 1st gear steep hill (towing iirc?) laws of the day with their 302 foxbodies as a cost-effective compromise with the size of tires/car they had. Is that relevant to most of us budget hacksters 25 years later?
Stock (diameter) 205-55-16 summer tires with the 2.95-.72 spread w/4.10s is quite painless/livable (of known TB-approved inexpensive 5-speed choices.) The SOHC motor can pull it in vacuum/lean burn cruise in 5th/OD (even with a (slight/minimal) load/trailer/roof rack load/~4-6kft interstate "high plateau" elevations/ hot thin air/less-than-ideal conditions) returning acceptable economy/NVH at the speeds possible/plausible with these cars (65mph/ less many times, whatever the drivetrain.)
Rev drop from 6K shifts ~1500RPM even with mild turbo cam(s) with easily street-able clutch. Boost drop 'tolerable' (subjective) with the closing of the throttle/inevitable power-flow-interrupt compromise of a manual transmission.
~29mpg + or - 2 frequently observed at cruise in many conditions (75-85+mph, beverly hillbilly roof rack (and its probable bulky contents), very long/steep hills/6K+Ft elevation, hot days and/or all of the above, no). Noise not discernibly worse than the dated boxy car with a bunch of glass/rear eco-chamber in the wind, (often factory) misaligned body panels/dried out seals/sound deadening/interior plastic/tractor vibrating 4cylinder ever is. Failure rate of the trans was uncommon(ish) with only 2.1-2.5L 4cylinder power/3000lbs/205 summer tires on the street or some track/auto-x/towing "mixed use" without constant abuse.
Fairly effective? recombining the used Volvo parts bin to use a 4.10 1041 gear set w/ Alloy diff cover w/ OE filter & magnet (Alloy cover conducts heat away more effectively, but can/will crack instead of dent unlike stamped steel cover if fed a rock diet...
...longevity/driving predictability *may* be impacted...
Slightly modded "almost always locked" G80 (many limitations notwithstanding/not exactly "driver/performance oriented" and/or strong (less predictable than welded or clutch LSD, but doesn't eat your tires/suspension/break axles either), more traction than open, & (probably all-importantly) TB-priced) coupled with various "best of" the OE T5 bits.
FWIW:
-Closest USA-market junkyard- able ratio spread T5.72OD x 4.10=2.952
-M46/47/Cosworth T5.80OD x 3.54=2.832 (OE on 700T stick shift cars) Pretty good all around street driver compromise with 205-55-16 OE tires.
-"".80OD x 3.73 = 2.984 (OE 240T) (a little spun out, especially with OE 195-60-15 turbo size tires, but "responsive" (tolerable?) with a 7.5:1 compression B21FT that was marketed as a "driver's car" in the day (80s Turbo cars typically got horrible test drive magazine reviews in the day new and manufacturers knew it).
-"".80 OD x 3.31= 2.648 (N/A 8V 2.3L fuel econ/late emissions era USA cars)...faints in the sight of hills...terrible/requires boost (in boosted SOHC redblock apps) negating any/all performance & economy benefits at all but ideal unloaded 85-100+mph in the flat.