I’ve always been a Doubting Thomas when it comes to turnaround talk at General Motors.

But I’ve seen enough evidence to say it: The auto business turnaround is real, despite the $39 billion third-quarter loss, which is the result of arcane accounting rules. More important is that production capacity has been reduced, and market share has leveled.

After the first six months of this year, GM’s market share was 23%, a sharp dive from 24.5% last year. Then in July it moved to 24.1%, in August to 26.1%, in September to 25.5% and in October to 25%: Four months of share increases over the six months before.

There always are explanations such as getting rid of old stock and rollouts of new models, but GM’s share has been creeping upwards even though marketing chief Mark LaNeve is holding down incentives.

The number of first-rate cars and trucks in GM’s portfolio also is growing. The stable includes big pickups and SUVs; the new cross/utility vehicles such as the Buick Enclave, Saturn Outlook and GMC Acadia, plus cars such as the Cadillac CTS, Chevy Malibu and Saturn Aura.

They call that a turnaround. It’s happening.

How far will it go? Alas, not as far as many of us hoped. I don’t expect to live to see GM at 30% of the market again. It’s reasonable to figure that GM will peak at 25%-27% of the market over the next few years.

The company policy is to back away from rental car fleets and giveaway prices to employees and neighbors of employees, to sit hard on the rebates and push for honest retail sales.

It is not going for volume even on the new stuff. Take the Malibu: From all the talk, you would think GM is out to catch the Toyota Camry or Honda Accord. Not true. In 2005, GM sold 203,500 of those dog-eared Malibus, well behind the Japanese models. It probably won’t sell 203,000 next year.

The difference is GM wants to make money, and not give them away. Look at the production setup: One plant in Kansas also makes the Saturn Aura, which is selling at 60,000-plus units a year. Two full shifts with heavy overtime might get 250,000 units of both cars.

In addition, from all the plant closings and layoffs, GM seems to be positioning itself for 22%-23% of the market. If it can hang on to 25%, without giving away the merchandise, it should be profitable.

Nevertheless, even these predictions depend on a few other successes:

The present globalization will have to actually work, and Americans will have to prove they are willing to accept cars designed and engineered by GM in Germany, Korea or Australia.

And GM Powertrain must keep up with the engine and transmission developments by the competition. I haven’t seen real evidence of that yet. The 6-speed automatic for the Malibu 4-cyl. isn’t in production yet, so the volume models of this new car carry a tired 4-speed auto while Honda and Toyota competitors have 5-speeds.

And hope the plug-in hybrid-electric car works, because the blow to GM’s credibility will be great if it can’t deliver in 2010 or 2011.

But why look for trouble? There’s good news tonight. I can see the yellow brick road ahead.

Jerry Flint is a columnist for, and former senior editor of, Forbes magazine.