delivered the opinion of the court:
Following a bench trial, the defendant, Charles A. Campbell, was convicted of residential burglary (Ill. Rev. Stat. 1989, ch. 38, par. 19— 3). The trial court sentenced him to serve five years in prison. He appeals, contending that he was not proved guilty beyond a reasonable doubt.
Jeffrey Miller testified at the defendant’s trial that on March 9, 1989, he resided at 308 Willard Street in Joliet with Susan Lewis Buchanan. Upon returning home that evening, Miller discovered that the lights were on and his house was in disarray. He noticed that there were muddy footprints throughout his living room and kitchen. Bills which had been on the kitchen table when he left that morning were scattered on the floor. He also noticed that a television and a VCR were missing. Upon examining the outside of his house, he observed footprints in the mud beneath the windows. On the day in question, Miller had not given anyone permission to enter his house.
Joliet police department evidence technician Rex Provensale testi*593fied that he conducted an investigation at the crime scene. On the floor, he found a telephone bill envelope bearing a shoeprint. Proven-sale examined the shoes of Miller and the other officers at the scene and determined that they did not match the print on the envelope. Provensale also retrieved a latent fingerprint from a videotape container. It was later established that the fingerprint did not match any of the defendant’s fingerprints.
Police officer Richard Fonck testified that on March 12, 1989, the defendant was at the Joliet police station on an unrelated matter. After observing that the defendant was wearing tennis shoes, Officer Fonck compared a photograph of the shoeprint on the envelope taken by Officer Provensale with the soles of the defendant’s shoes. Based on that examination, Fonck concluded that the designs were similar and confiscated the defendant’s shoes. Thereafter, the shoes were sent to the State crime lab for examination.
Walter Sherk, a forensic scientist, testified that he worked in the area of firearms, tool marks, and shoeprints. He noted that he had taken courses and attended various lectures on shoeprint analysis. In his 14V2 years as a forensic scientist, he had performed approximately 300 shoeprint comparisons and had testified in 15 cases about his conclusions from shoeprint comparisons. Sherk further testified that he had made a comparison of the Nike brand tennis shoes taken from the defendant and the shoeprints found on the envelope at the scene of the burglary. He noted that the envelope bore two separate shoe impressions. Since the patterns were dissimilar, he concluded that the smaller of the two prints on the envelope could not have been made by the defendant’s shoe. Sherk then made a test print from the defendant’s right shoe for a comparison with the larger of the two prints on the envelope. The larger impression consisted of two-thirds of the middle portion of a shoe. Based on his comparison, Sherk found that the sizes of the shoes and the patterns on the soles were consistent. Sherk then compared the shoeprints for similar individual characteristics. He explained that individual characteristics were nicks, cuts, and scratches that appeared in the pattern of a shoe over a period of time as a shoe was worn. Sherk’s comparison revealed six similar individual characteristics. Based on the matching characteristics, he concluded that the larger print on the envelope had been made by the defendant’s right shoe.
On cross-examination, Sherk acknowledged that there was not a set number of individual characteristics that had to be similar to identify a shoe. He stated that as little as two or three similar marks might be enough. Sherk admitted that wearing shoes for two or three *594days could cause some change in the characteristics of the soles. He also noted that he had not looked for dissimilarities and that dissimilarities could affect the comparison. He further stated, however, that given a dissimilarity, he would assume that the shoe was distorted after the original impression was made or that dirt was present at the time the impression was made.
Police officer James Klancher testified that while on patrol on May 24, 1989, he had observed the defendant walking along a residential street. The officer informed the defendant that he had a warrant for his arrest. At that point, the defendant fled. The officer gave chase and finally apprehended him at his residence.
Nineteen-year-old Tana Ramos testified on behalf of the defendant that on or about March 9, 1989, she resided at 306 Willard Street in Joliet, Illinois, which was next door to the burglarized home. She was a friend of the defendant and had known him since she was in the sixth grade. She stated that the defendant came to visit her almost daily.
The defendant argues on appeal that his conviction should be reversed because the shoeprint evidence, standing alone, was insufficient to establish his guilt beyond a reasonable doubt.
It is well settled that, on review of a conviction, this court must examine all the evidence in the light most favorable to the State to determine whether any rational trier of fact could have found the defendant guilty beyond a reasonable doubt. (People v. Collins (1985), 106 Ill. 2d 237, 478 N.E.2d 267; People v. Pintos (1989), 133 Ill. 2d 286, 549 N.E.2d 344.) A criminal conviction will not be set aside unless the evidence is so improbable or unsatisfactory that it creates a reasonable doubt of the defendant’s guilt. People v. Vriner (1978), 74 Ill. 2d 329, 385 N.E.2d 671.
After reviewing the evidence in the light most favorable to the prosecution, we conclude that the defendant’s conviction must be reversed. The only evidence connecting the defendant to the crime was the shoeprint evidence. Although such evidence may be used in conjunction with other circumstantial evidence to support a conviction (see People v. Diaz (1988), 169 Ill. App. 3d 66, 522 N.E.2d 1386; People v. Stanbeary (1970), 126 Ill. App. 2d 244, 261 N.E.2d 765), we are unaware of any Illinois precedent establishing that such evidence alone can support a conviction. In this regard, we note from the expert’s testimony that shoeprint analysis does not appear to have the exacting standards applied to fingerprint analysis.
Further, we find the State’s sole reliance on the shoeprint evidence problematic in this particular case given the questionable relia*595bility of that evidence. The State’s expert admitted that he did not check for dissimilarities. It was unclear from his testimony whether such dissimilarities would affect his analysis. Additionally, it was established that three days’ wear could change the contour of a shoe’s sole and that the instant defendant could have worn the shoe in question for up to three days after the crime. We also note that no evidence was presented establishing that the defendant was wearing the shoe in question on the date of the offense. Under the circumstances, we hold that the State’s evidence left a reasonable doubt about the defendant’s guilt.
The judgment of the circuit court of Will County is reversed.
Reversed.
BARRY, J., concurs.