DNA tests have furnished a double surprise for Thomas R. Robinson, an associate professor of accounting at the University of Miami. The first was being told he was descended from Genghis Khan. The second was learning last week that the first test was wrong.

Mr. Robinson's own caution was one cause of this vicissitude. His temporary induction into the Mongol royal house began in April, when he received a call from Oxford Ancestors, an English DNA testing company he had asked in 2003 to test his Y chromosome. The company said that on a recent scan of its database Mr. Robinson's chromosome had emerged as having a genetic signature very close to Genghis Khan's.

Mr. Robinson's male ancestors were British. Perhaps because of that apparent incongruity -- and the contrast between his profession and Genghis Khan's -- the finding was reported in several newspapers, including The New York Times.

Then a movie company offered to fly Mr. Robinson out to Mongolia. But instead of sitting back and basking in the posthumous fame of his new ancestor as many people might have done, Mr. Robinson decided to get a second opinion before matters went any further.

He sought the view of Family Tree DNA of Houston, only to learn last week from its president, Bennett Greenspan, that he belonged to a different branch of the Y chromosome family tree from that of the Mongol emperor, and could not be descended from him.

Chris Tyler-Smith, a geneticist at the Sanger Institute in Cambridge, England, said Family Tree DNA's analysis was correct. The company's test of Mr. Robinson's Y chromosome ''conclusively rules out a link to the Genghis Khan haplotype,'' he wrote in an e-mail message.

A team led by Dr. Tyler-Smith identified the haplotype, or genetic signature, of the Mongol royal house in 2003 after a survey showed it to be carried by an estimated 16 million men living in the lands of the former Mongol empire. The domain of Genghis and his heirs stretched from the Caspian Sea to the Pacific Ocean, and Genghis is said by contemporary historians to have labored assiduously in his large harems.

Mr. Robinson said that he was disappointed only that Oxford Ancestors had not run the same test as Family Tree DNA had done, or at least told him it was available.

Bryan Sykes, the founder of Oxford Ancestors, said he agreed with Dr. Tyler-Smith's verdict but maintained that his original interpretation was reasonable.

The discrepancy occurred because Oxford Ancestors only tested Mr. Robinson's Y chromosome at nine sites, ones at which the DNA mutates quite often between generations. Finding a match between Mr. Robinson and Genghis at seven of nine sites, Dr. Sykes assumed that was good enough to declare a direct relationship, since he had never seen such a match outside of Asia, he said.

But the major branches of the Y chromosome family tree are defined by mutations at sites that change very seldom. Oxford Ancestors did not check the slow-mutating site that defines the branch to which Genghis Khan belongs. But at Family Tree DNA, Mr. Greenspan saw a feature in one of the fast-mutating DNA sites that bothered him, and did the costlier test of a slow-mutating site.

A match at 10 fast-mutating sites is outvoted by a discrepancy at one slow-mutating site, Dr. Tyler-Smith said.