JOSEPH JEROME RAMIREZ, APPELLANT,

v.

STATE OF FLORIDA, APPELLEE

OVERTON, Justice.

Ramirez appeals his conviction of first-degree murder and his sentence of death. We have jurisdiction pursuant to article V, section 3(b)(1), of the Florida Constitution. This is Ramirez’s second appeal to this Court. In his first appeal, this Court reversed Ramirez’s murder conviction and vacated the death sentence imposed at his first trial. We did so because the State failed to establish a sufficient predicate for its expert’s assertion that Ramirez’s knife was the only knife in the world that could have caused the victim’s wounds. Ramirez v. State (Ramirez I), 542 So. 2d 352 (Fla.1989).

In the instant case, the State endeavored to comply with our instructions in Ramirez I by attempting to establish the reliability of the scientific tests supporting the expert’s assertion at a hearing conducted immediately preceding Ramirez’s second trial. Because we find that Ramirez was denied his due process right to present evidence refuting the State’s evidence of reliability at the pretrial hearing, we must once again reverse his conviction.

The relevant facts surrounding this murder are reported in our previous opinion. See Ramirez, 542 So. 2d at 352-54. Testimony at the first trial revealed that the murder victim had been stabbed twelve times and beaten in the head with a heavy object. At that trial, the State introduced into evidence a knife linked to Ramirez. Thereafter, an expert offered his opinion that, based on his scientific examination and comparison of a cast made from Ramirez’s knife and a cast made from knife marks found on the victim’s cartilage, Ramirez’s knife was the only knife in the world that could have been used in the murder.

On appeal, this Court stated: “[W]e find that no scientific predicate was established from independent evidence to show that a specific knife can be identified from the marks made on cartilage.” 542 So. 2d at 354-55. We found that the error in admitting the expert’s testimony was not harmless and remanded for a new trial.

The record of the instant case indicates that the State requested a special hearing before Ramirez’s retrial to present testimony and evidence to the trial judge relating to the reliability of knife-mark comparison evidence and its admissibility in the upcoming trial. The judge granted this request and the State presented documentary evidence as well as depositions and live testimony concerning the theory, practice, and procedures involved in knife-mark comparisons. Much of the testimony was challenged by the defense through vigorous cross-examination. At the close of the State’s presentation at the pretrial hearing, defense counsel proffered an expert to testify against the scientific reliability of knife-mark comparisons. The trial judge refused to allow the defense to call the witness. The trial judge stated that any evidence presented by the defense to refute the reliability of knife mark-comparisons could be presented to the jury as the finder of fact, but such evidence was simply not relevant to the issue of basic admissibility.1 Ramirez has challenged both the admissibility of this evidence and the trial judge’s refusal to allow him to present his own expert at the admissibility hearing.

The admission into evidence of expert opinion testimony concerning a new or novel scientific principle is a four-step process. See generally Charles W. Ehrhardt, Florida Evidence § 702.1 (1992 Edition); Michael H. Graham, Handbook of Florida Evidence § 90.702 (1987 Edition).

First, the trial judge must determine whether such expert testimony will assist the jury in understanding the evidence or in determining a fact in issue. § 90.702, Fla.Stat. (1993) (adopted by the Florida Supreme Court in In re Florida Evidence Code, 372 So. 2d 1369 (Fla.1979)).

Second, the trial judge must decide whether the expert’s testimony is based on a scientific principle or discovery that is “sufficiently established to have gained general acceptance in the particular field in which it belongs.” Frye v. United States, 293 F. 1013, 1014 (D.C.Cir.1923).

This standard, commonly referred to as the “Frye test,” was expressly adopted by this Court in Bundy v. State, 471 So. 2d 9, 18 (Fla.1985), cert. denied, 479 U.S. 894, 107 S.Ct. 295, 93 L.Ed.2d 269 (1986), and Stokes v. State, 548 So. 2d 188, 195 (Fla.1989).

The third step in the process is for the trial judge to determine whether a particular witness is qualified as an expert to present opinion testimony on the subject in issue. § 90.702, Fla.Stat. (1993).

All three of these initial steps are decisions to be made by the trial judge alone. See Johnson v. State, 393 So. 2d 1069, 1072 (Fla.1980), cert. denied, 454 U.S. 882, 102 S.Ct. 364, 70 L.Ed.2d 191 (1981); Rose v. State, 506 So. 2d 467 (Fla. 1st DCA), review denied, 513 So. 2d 1063 (Fla.1987).

Fourth, the judge may then allow the expert to render an opinion on the subject of his or her expertise, and it is then up to the jury to determine the credibility of the expert’s opinion, which it may either accept or reject. Wuornos v. State, 644 So. 2d 1000, 1010 (Fla.1994) (“[T]he finder of fact is not necessarily required to accept [expert] testimony.”); Walls v. State, 641 So. 2d 381, 390 (Fla.1994) (“[E]xpert opinion testimony [is] not necessarily binding even if uncontroverted.”).

The second step, concerning whether to allow expert opinion testimony on a new or novel subject, is especially important to the process. As Professor Ehrhardt has explained:

When a novel type of opinion is offered, the proffering party must demonstrate the requirements of scientific acceptance and reliability. The most widely adopted test has been that of Frye v. United States which involved the admissibility of an early polygraph. The court held the evidence inadmissible because the underlying scientific principle was not “sufficiently established to have gained general acceptance in the particular field in which it belongs.”

Ehrhardt, supra, § 702.2 (footnotes omitted).2 The principal inquiry under the Frye test is whether the scientific theory or discovery from which an expert derives an opinion is reliable. We have not hesitated to utilize the Frye test to reject expert testimony concerning subjects that have not been proven to be sufficiently reliable. See, e.g., Ramos v. State, 496 So. 2d 121, 123 (Fla.1986) (testimony of dog-handler and police officer insufficient, by itself, to establish reliability of dog scent-discrimination lineups); Bundy v. State, 471 So. 2d 9, 18 (Fla.1985) (hypnotically refreshed testimony per se inadmissible), cert. denied, 479 U.S. 894, 107 S.Ct. 295, 93 L.Ed.2d 269 (1986), modified Morgan v. State, 537 So. 2d 973 (Fla.1989) (defendant’s refreshed testimony may be admissible); Walsh v. State, 418 So. 2d 1000, 1002 (Fla.1982) (“[P]olygraph evidence is inadmissible in an adversary proceeding in this state.”); Zeigler v. State, 402 So. 2d 365, 373 (Fla.1981) (“The results of a sodium butathol test are not admissible in a criminal prosecution.”), cert. denied, 455 U.S. 1035, 102 S.Ct. 1739, 72 L.Ed.2d 153 (1982).

The district courts of appeal have likewise refused to permit unreliable expert testimony to reach the jury. In Copeland v. State, 566 So. 2d 856, 858 (Fla. 1st DCA 1990), the First District Court of Appeal found that the admission of expert testimony based on the “spermatozoa age test,” where there was not a proper predicate to establish the scientific reliability, was error, and noted that “the only evidence presented was of noncontrolled experimentation by the crime analyst.” In Page v. Zordan, 564 So. 2d 500, 502 (Fla. 2d DCA 1990), the Second District Court of Appeal held that it was error to permit an expert clinical psychologist to rely upon a “sexual abuse legitimacy scale” because no predicate was established concerning the acceptance of the test in the scientific community. In Crawford v. Shivashankar, 474 So. 2d 873, 876 n. 4 (Fla. 1st DCA 1985), the First District Court held that the trial court was not required to accept an expert’s “bald assertion” that thermography had received general acceptance in the relevant scientific community.

In utilizing the Frye test, the burden is on the proponent of the evidence to prove the general acceptance of both the underlying scientific principle and the testing procedures used to apply that principle to the facts of the case at hand. The trial judge has the sole responsibility to determine this question. The general acceptance under the Frye test must be established by a preponderance of the evidence.

Just as important as the burden of proof is the fact that the hearing must be conducted in a fair manner. There is no question that a hearing on the admissibility of novel scientific evidence is an adversarial proceeding in which conflicting evidence is presented to the trial judge as the trier of fact. Without the testimony of experts presented by both parties, the trial judge is denied a full presentation of relevant evidence. This is especially important in a criminal trial where the defendant is guaranteed certain constitutional rights, not the least of which is the due process right to present witnesses in one’s behalf.

In the first trial of this case, we held that the State had not established the reliability of knife-mark comparison evidence when the proof of reliability was limited to the expert technician’s statements that the technique was reliable together with an arti-ele the technician had written concerning the technique. Ramirez v. State, 542 So. 2d 352, 354-55 (Fla.1989).

In the instant case, the State presented substantially more evidence of reliability that, if believed by the trial court, may have established the reliability of knife-mark comparisons.3 However, it is impossible to determine whether the evidence presented by the State was sufficient to prove the reliability of knife-mark comparisons because Ramirez, despite his offer of proof, was denied the right to present any evidence to the contrary at the pretrial hearing.4 This was a clear violation of Ramirez’s due process rights. Without the presentation of Ramirez’s evidence concerning the lack of reliability of knife-mark comparisons, the trial judge’s determination that the evidence was reliable was error. We are unable to find that the error was harmless.

As we stated in our prior opinion in this case, the State is not precluded from introducing Ramirez’s knife into evidence and presenting testimony that the wounds on the victim were consistent with that knife. That is not how the State decided to proceed and this case must again be retried. We request the Chief Judge of the Eleventh Judicial Circuit to do everything possible to expedite the retrial of this matter so that Ramirez’s retrial will be commenced within 90 days from the date of mandate from this Court.

It is so ordered.

GRIMES, C.J., and SHAW, KOGAN, HARDING, WELLS and ANSTEAD, JJ., concur. . The record of the pretrial hearing indicates that the trial judge did not comprehend the need for testimony proffered by the defense concerning the lack of reliability of knife-mark comparisons. The trial judge made the following comments in refusing defense counsel's proffer of its own expert:

We do not have a mini trial. We have a unilateral hearing outside the presence of the juiy for the admission of the evidence, because all I rule on is its admissibility, not its weight. That will be left for the triers of fact, the jury, in this case. Now, I cannot comprehend what evidence the defense would have to offer that would be relevant to this issue. If you have got something to tell me that would make it relevant, tell me. But I don’t know of any, and I am not going [to] permit defense testimony at this time.

(Emphasis added.) As will be explained in this opinion, these statements reflect the trial judge's misunderstanding of the procedures involved in a hearing on the admissibility of novel scientific evidence.

. Professor Ehrhardt also notes that some Florida district courts of appeal had taken the position that section 90.403 of the Florida Evidence Code superseded the Frye test. Ehrhardt, supra, § 702.2. We clarified any confusion on this issue in Stokes where we noted that Florida continues to follow the Frye test. See also Flanagan v. State, 625 So. 2d 827 (Fla.1993).

. We expressly decline to comment on whether the State has presented sufficient evidence to establish the reliability of knife-mark comparisons.

. We approve the use of a pretrial hearing for the admissibility of evidence as was done in this case. While the trial judge may excuse the jury and conduct a hearing on admissibility during the trial, we find that the better procedure is to have the hearing before the trial begins. This will minimize any inconvenience to the jury. Cf. Cirack v. State, 201 So. 2d 706 (Fla.1967).