The triumvirates, the social wars, the third century crisis, etc. had tons of Romans killing one another. While many wars against foreigners often had relatively few Roman casualties. Is it probable that more Romans died fighting amongst themselves than against foreign powers? Is it even possible to know?
I've seen this question asked before but it has never been sufficiently answered, and I don't think it can be. The long story short is: we don't know.
Roman civil wars were massive, and in a span of 1900-2200 years of History (depending on what dates you pick) they did have more civil wars and usurpations than any other civilization, except maybe for China. The largest battle in Roman history is believed to be the Battle of Philippi - allegedly 223,000 men vs. 187,000 men. Most of Rome's civil wars were like this - Bedriacum (69), Adrianople (324), Mursa (353), etc.
However, the Romans spent far more time fighting incursions than they did themselves. The infamous battle of Cannae allegedly saw around 50,00-70,000 killed, according to Livy and Polybius respectively. Attila's devastating victory at the Battle of the River Utus is said to have destroyed three field armies - probably around 75,000 men, and a significant chunk of the Danubian Limes went with it, for example. (See Williams and Friell, "The Rome that Did not Fall," basically all of Chapter 7). The casualties from his invasion of the Balkans may have well rivaled some of Genghis Khan's numbers, but we really don't know.
In reality the number is going to be skewed by sourcing - sources for the Civil Wars of Rome are often far more detailed, particularly for the Principate and Late Republic, than they are for sources about how many Romans were slain in the rather unusual defeat surffered at the hands of barbarians. Adrianople, one of the most infamous, may have only seen 20,000 men slain. Yet also by comparison, most of the 187,000 man army at Phillipi surrendered, and was not slaughtered.
It's a question that can't properly be answered, although someone with enough time and dedication might be able to come up with a rough guess based on what is known. I know of no author who has ever attempted to do so.