The rape case hinges, as often happens, on one person's word against another's.

The 14-year-old girl told D.C. police last April she had been at home alone a day earlier, when her sister's boyfriend forced her to have sex. She had evidence: a semen-stained pair of panties she said were hers.

With the aid of DNA profiling -- a revolutionary forensic technique accepted by courts in 38 states, including Maryland and Virginia -- reading the genetic code of that stain might be all it takes to send Kevin Porter, 22, to jail for at least 15 years or to clear him. He is charged with attempted carnal knowledge of a minor -- the equivalent in the District of statutory rape. He has pleaded not guilty.

Porter's attorney is mounting a defense that questions the reliability of such DNA profiling. This counterattack does not dispute the technique's underlying principle -- that each person's DNA (deoxyribonucleic acid) is unique -- but focuses instead on the absence of any uniform national standards governing the way the test itself is performed.

"Here's a guy who's facing 15 years to life, and anybody who wants to can set up a DNA testing lab in his garage," attorney Ronald A. Goodbread said during a break in a recent hearing in the case. "When you go in for a Pap smear, there are generally accepted {laboratory} protocols. . . . Hell, when you go get your hair colored, there's a state licensing board for cosmetology."

But there are no generally accepted standards for labs that do DNA profiles -- and there is a growing number of private and state labs eager to take advantage of what promises to be the most important forensic tool since fingerprints. The technique is capable of identifying suspects with a certainty that far exceeds eyewitness testimony.

Since its first use in a British murder case in 1987, DNA profile evidence has been offered in 45 states and at least 15 foreign countries, according to an informal survey done last year by the U.S. Office of Technology Assessment (OTA). Courts in several of those states, however, including the highest court in Massachusetts last month, have backed away from allowing the tests to be used as evidence against criminal defendants, citing concerns over their reliability.

If it is admitted in Porter's case, it would mark the first time a DNA profile has been accepted as evidence in a District case; earlier cases involving DNA evidence here ended up not going to trial.

Few doubt any more that advances in the study of human genetics make it possible, in theory, to examine the genetic components of human cells from blood, semen, bits of hair, skin or bone left behind at a crime scene by the perpetrator and compare them to the DNA in samples taken from a suspect. And, in theory, a study of those codes should reveal if there is a "match" -- almost certain proof that the suspect was at the scene of the crime.

But the test's potential for error lies behind defense attorneys' newest challenge to the testing procedure, a challenge in which they have found themselves in a limited but unlikely alliance with the scientific community.

In a report issued last July, the OTA said there was an "urgent" need to develop a national consensus on how to perform DNA profiles -- so that all analyses use the same basic technology, chemical reagents, criteria for declaring a "match." The report said it was also important to develop a uniform approach to the statistics used to interpret the chances of a match being an accident. "Standards are necessary if high-quality DNA forensic analysis is to be ensured," the report said, "and the situation demands immediate attention."

The absence of generally accepted standards for the tests stems from the newness of the technique, which has led to constant refinements of the technology, and the multiplicity of laboratories pioneering the tests.

As a result, the three major laboratories that do the most common type of DNA profiling -- two private companies, Cellmark Diagnostics of Germantown, Md., and Lifecodes Corp. of Valhalla, N.Y., as well as the FBI -- each use different chemical techniques for the process and different sets of statistics with which to interpret their results.

John W. Hicks, director of the FBI's forensic laboratory, said he thinks the FBI's method will become the national standard used by state crime labs, leaving private laboratories such as Lifecodes and Cellmark free to develop the market for DNA profiling in civil cases such as paternity disputes.

But until there are national standards, defense lawyers say they fear that different techniques make it possible for one expert to see a match where one does not exist, or to make a mistake in calculating the odds that a match might be accidental. Their objections center on the level of uncertainty involved in obtaining a genetic profile.

Although actual procedures differ from lab to lab, they are similar because they are based on the same scientific concepts.

A lab takes the human cells from a crime scene -- often semen or blood -- and extracts the DNA from their nuclei. The DNA, the molecule of which genes are made, looks like a twisted ladder whose rungs are composed of four chemical units that always bond horizontally in pairs: adenine to thymine to make one rung, guanine to cytosine to make another. Along either side of the ladder, the four units can be linked in any sequence.

The DNA ladder is then cut into pieces with enzymes coded to look for specific points along the sides of the ladder. The enzyme used in the FBI procedure, for example, severs the DNA strand every time it sees the sequence guanine-guanine-cytosine-cytosine. Because the number and location of those sequences are the same in all of a person's cells but vary from person to person, the number and size of the DNA fragments will, too.

The negatively charged DNA fragments are then placed into an electrically charged gel to sort them by size. The fragments move toward the positive pole at varying speeds, depending on how big they are. After a period of time, the fragments become spread through the gel in a line, like race horses in a mid-race snapshot. The DNA fragments are spread according to their size.

The fragments are then blotted onto a nylon membrane and treated with "probes," which are synthetic bits of radioactive DNA with known sequences.

The probes bind only to complementary sequences in the blotted fragments and any unbound probe material is washed away. The membrane is then exposed to X-ray film, which produces an image wherever the radioactive probes have bound. What emerges from the developed film is something that looks like a blurry enlargement of the bar codes found on many consumer products -- a series of bands in various patterns corresponding to different points of an individual's DNA strand. This is the genetic "fingerprint."

Each person's cells produce a unique genetic "bar code." If the sequence of the bars from evidence at the crime scene exactly matches the bars produced from a suspect's cells, detectives conclude that only the suspect could have left the evidence at the crime scene.

Differences in the way labs do their tests have to do with the type of enzymes and DNA probes used. In other words, a person's genetic fingerprint will look different depending on which lab did the test.

But defense attorneys point out that reading those bar codes is not as simple as prosecutors may make it sound. Sometimes DNA from evidence samples migrates through the gel faster than DNA from blood taken in a lab, a phenomenon known as "bandshifting," which could produce a false match or a false mismatch. Sometimes, evidence from crime scenes produces pictures in which some bars are too faint or too blurry to read. One expert might see a match where another sees an uninterpretable picture.

More controversy centers on the use of population genetics to calculate the odds that any given match might be accidental.

Kevin Porter's case is typical. The FBI lab that analyzed the semen stain found a match between it and Porter's genetic profile, and calculated that the odds of it being an accidental match were roughly 1 in 30 million.

But where do the numbers used to calculate those odds come from? The FBI, Cellmark and Lifecodes all have developed their own databases of genetic information from various ethnic groups -- all of which have the same basic human genetic components, but in varying frequencies. Cellmark and Lifecodes buy their data from blood banks around the country, and also keep their own records; part of the FBI database was developed from blood samples submitted by cooperative FBI agents.

Defense lawyers say more research is needed to make sure that the statistical probabilities being calculated from these disparate sources are not wildly inaccurate. In Porter's case, for example, attorney Goodbread cited five cases in which experts had to recalculate the statistical odds because of an earlier mistake. The differences between the original estimates and the later ones were vast, in one case starting out at 1 in 14 billion and winding up at 1 in 140 million.

Scientists say that as disparate as those two calculations are, it's a distinction without a difference; odds of one in 140 million still constitute overwhelming proof.

To Hicks, the statistical arguments mounted by defense lawyers are red herrings. While DNA profiling needs refinements, he said, there is proof it works and that the growing adoption by state crime labs of the FBI's methods demonstrate that the criminal justice community is already on its way to establishing a uniform way of doing the test.

Spokesmen for Cellmark and Lifecodes said they also see the need for uniformity -- both to protect innocent defendants and to bolster their own credibility. Lifecodes has begun to offer the FBI testing method in addition to its own.

"We do not want the quality of work done by Cellmark after 2 1/2 years of testimony to be diluted by any lab which is doing less than competent work," said Mark Stolorow, manager of forensic services for Cellmark. With ongoing research will come more agreement on methods, he said, and "eventually this issue will go away."