The Cadillac CTS Out-Drinks the BMW 3 Series. It Still Kills at Half the Rate.

According to the toxicology reports, and there are a lot of them, Cadillac's great BMW-fighter has a drinking problem. Of 931 CTS drivers in fatal crashes with a tox screen on file, 25.9 percent tested positive for alcohol or drugs.^[1] Now run the division rivals. BMW 3 Series: 22.1 percent. Mercedes C-Class: 21.8 percent. Audi A4: 21.6 percent. Lexus IS: 21.4 percent. Cadillac spent the better part of two decades, and a sum of money nobody will admit to, building a German sports sedan out of Detroit parts. It worked. Its customers brought the German sports sedan's worst habit with them, then improved on it.

Start with the box score, because the box score is the whole column. That lead over the 3 Series is 3.8 percentage points with a standard error of about 1.7, which puts it at z = 2.29. That clears the bar. This is a real lead on a 931-driver sample, and it is overwhelmingly an alcohol story: 20.6 percent of CTS drivers were alcohol-positive against 17.5 for the BMW, while the drug gap is a modest 10.2 to 8.7.^[1] It is also a family scandal. Across the rest of Cadillac's lineup, the weighted average is 20.2 percent impaired. Its little brother the ATS, the other BMW-fighter, sits at 20.6 percent. XTS land yacht: 17.9 percent. Whatever is happening with the CTS, it is specific to the CTS, and it is not the badge. It is the car.

Now the twist, because every good box score has one. Flip from the toxicology table to the actual dying and the story inverts completely. CTS fatality rate: 1.32 deaths per 100 million vehicle miles. 3 Series: 2.73.^[1] Read that again. And yet the car with the drunker drivers kills at less than half the rate. If impairment explained fatality rates, the lazy assumption this column exists to audit, the CTS would be the bloodier car by a wide margin. Instead the engineering eats the entire impairment gap and asks for seconds. Chassis: exonerated. Drivers: not.

Now the counterargument, stated at full strength, because the numbers invite a cleaner story than they deserve. Impairment here is measured only among drivers already in fatal crashes, so "CTS drivers are drunker" overstates it; the precise claim is that CTS drivers in fatal crashes test impaired more often. In the FARS window, the CTS fleet skews older and cheaper than the 3 Series fleet, since a 2005 CTS was a six-thousand-dollar used car by 2018, and cheap old performance sedans select for exactly the risk-tolerant young men who drink and drive. All of this is nameplate-level, which folds three CTS generations and the supercharged CTS-V halo car into one number. And the rate gap favoring the CTS may reflect exposure differences in the estimated miles rather than superior crashworthiness.^[2] All true. None of it erases a 3.8-point lead at z = 2.29.

What survives is the part that should annoy everyone with a simple theory. Cadillac's fighter was rear-drive, offered a manual, spawned the CTS-V, and did everything asked of it except attract sober buyers. Its drivers test impaired at a higher rate than the Germans it hunted, and the car still buries the BMW on the only table that counts, the one measured in bodies per mile. Two lessons, both usable. If you are shopping a used CTS, price the insurance before you fall in love with the exhaust note; the car's engineering is not the problem and the nameplate's driver pool is. And the next time a fatality rate looks bad, check the toxicology table before blaming the chassis. Sometimes the chassis is guilty. This time it has an alibi, signed by the morgue.