When you paid your Visa or MasterCard bill last month, a computer somewhere took note.

And there is a good chance that the computer not only took note of your payment, it gave it some thought. Like whether you paid the bill on time and maybe even what merchants you dealt with.

The reason it did all this thinking is simple: The bank that issued you your credit card is a lot more interested in you than it used to be.

With the economy slumping and the credit card market saturated, card issuers are taking a hard look at their customers both to try to head off problems before they grow large and to keep good customers from going elsewhere.

In the past, "credit card portfolios have been a source of substantial revenues for banks ... by continuing to grow," said Jim Roemmer of Credit Partners, a banking computer services firm based in Larchmont, N.Y.

Now, though, growth is getting hard to come by and solicitations have lost their force, making the need to identify and keep the best customers all the more critical.

And to assist the banks in this, there are more powerful computers and more sophisticated software that allow them to collect and retain more information and sort through it in finer detail than ever before.

Though civil libertarians and others concerned with privacy issues worry about the potential for disseminating data to people who might abuse it, the banks regard these efforts as a key tool to remaining profitable in an intensely competitive environment.

In many ways, bankers add, the new computer systems simply make the same kinds of judgments that were once made by people. Most are based on "scoring" models, which assign a customer scores in certain categories, such as payment history, total debt and the like, and then use those scores to decide whether to grant a customer credit and how much.

"The tool is such now that if you are of any size and sophistication, in order to be competitive you've got to use scoring systems," said Roemmer.

But he added that "there's nothing really Buck Rogers about it. It's just that computational ability of the data-processing software is such that more sophisticated determinations can be made. The basic premise on how to design a scoring system -- that's not changing as rapidly."

Buck Rogers or not, the new systems allow bankers to make increasingly sophisticated predictions about whether their card holders will pay their bills.

Bankers and others say that an individual's own credit history is by far the most reliable predictor, while generic data such as demographics are far less dependable and can raise discrimination questions.

And the parameters are being adjusted to reflect current conditions.

For example, a person with multiple credit cards, as long as he or she kept them all current, counted favorably on older scoring systems. But now some issuers are changing that.

One Southern banker noted that small-business owners he knows often take every card solicitation they get, regarding the extra cards as a backup source of cash in case of disaster. "That's not what we want," he said.

Another vendor said there are three major areas where the software is used today:

Predicting whether a newly delinquent account will return to "current" status or become seriously delinquent and possibly default. Such a forecast helps the bank decide whether to mount a prompt collection effort, let the account ride or step in and cut the consumer off entirely.

Deciding whether to grant an increase in a customer's credit limit. If the account is current, the software can predict whether it will remain so even if more credit is extended.

Evaluating over-the-limit authorizations. If a card holder seeks to charge more than he or she is currently allowed, the system, based on its scoring model, can authorize the additional credit automatically or kick the request out, along with the score, to a human evaluator. This usually involves an item large enough for the merchant to check before accepting the card.

In addition, banks also may call upon their systems for help in deciding whether to reissue a card to someone with a less-than-perfect payment history.

In practice, said Darryl Hansen, president of Norwest Card Services in Des Moines,this means that "the software goes in and looks at the attributes of a customer's account and how that account is being handled {by the customer}. And through a statistical methodology it can predict whether that account might have high probability of becoming delinquent or a low probability of becoming delinquent.

"It looks at payment patterns, usage patterns ... " Hansen added. "We look at every account automatically and assign a behavior score. That score is, if you will, an odds quote, 100 to 1 or 1,000 to 1.

"It can't predict a specific account -- it can't look at Darryl Hansen and say that definitely someday he will be delinquent. It can look at a lot of accounts and predict that out of that pool of, say, 10,000 accounts that one of them will go delinquent. So we can look at that pool and get a sense of risk from the profile."

Norwest, with 2 million accounts, uses the software not only as a means of protecting itself from defaults but to try to improve service, sometimes in ways the consumer is not even aware of, Hansen said.

For example, if an over-the-limit transaction is approved automatically, the customer may know only that he or she didn't have to spend very long at the cash register.

On the other hand, "if a customer's score is very low, suggesting high-risk profile, then a transaction above the limit might not be authorized" at all, Hansen said.

The software "helps us to discriminate, to be more selective in servicing those customers that are handling their accounts properly," he said.

In addition, instead of treating all 2 million customers as one account, they can be sorted out to solicitations for additional bank services. This not only cuts the costs of such solicitations but also spares those least likely to want the service from receiving such mail.

Hansen said his bank is still learning how best to use automation. "There's a lot of technology out there, but not all of it is right," he said, adding that there is "a learning curve" that he and other card issuers continue to climb.

The technology could be abused, Hansen said. Security could be breached or data misused, he conceded, "but that is obviously not in our best interests" and "we work very hard at preventing that from happening."