From Wikipedia:

During the 1st and 2nd centuries, the Roman Empire had a population estimated in the range of 59 to 76 million.[2] The population likely peaked just before the Antonine Plague. Harper[3] provides an estimate of a population of 75 million and a population density of about 20 people per square kilometre during its peak. In contrast to other ancient and medieval societies, the Roman Empire appears to have had unusually high urbanization rates. During the 1st and 2nd centuries CE, the population of the city of Rome is conventionally estimated at one million inhabitants. Ian Morris[4] estimated that no other city in Western Eurasia would have as many again until the 19th century.

There is a explanation, maybe? The Urban density was rural farmers moving to cities after the agricultural owning elite kicked them off their land and was using slave labor to grow crops.

One of Romes greatest conundrums was figuring out who to fuck over with rural/agricultural reforms. Do they screw over the rich and give it back to the poor? Or do they let the rich land owner continue to work these fields. That is a VERY small summary of the situation regarding this but I think that may be a root cause, at least in the Italian peninsula.

That urbanisation estimate is deeply flawed (the wikipedia entry even claims a higher percentage than the source it cites): don't believe everything you read, least of all on wikipedia. The claim that Roman urbanisation was "twice as high as that of Europe at the turn of the 19th century" (an assertion likewise not made by the cited authors) rests on combining a thus inflated Roman percentage (with places as small as 1,000 inhabitants reckoned as "urban") with a far more conservative one for later Europe (here probably with a lower limit of 5,000): applying a common definition they seem about evenly matched, with Europe c.1800 probably slightly ahead - itself a remarkable finding: Rome in this respect needs no exaggeration.

The claimed million for the ancient city of Rome is itself problematical: there are sources indicating such a total, but there's also a good case for a lower figure on the basis of area and likely densities. These estimates are far from settled. Beware the lure of big-looking numbers.