Shortly after I placed a six-pack of beer—Ska Brewing’s Modus Hoperandi IPA, for the hopheads out there, at 6.8% ABV—on the counter, I took a deep breath and exhaled slowly.
Then I did it again. And again. And again.
After each exhalation I watched as the numbers “0.000” appeared on my smartphone’s screen. My blood alcohol concentration, or BAC, was zero—just as I expected. With a bottle opener in one hand and a cold brew in the other, I removed the bottle’s cap. Crck-fsssst! Ah, the things I must do to bring you this weekly column.
In seriousness, though, that all-too-familiar sound kicked off another session of technology testing. For the rest of the night after finishing my drink, I waited the required 20 minutes to ensure any alcohol residue in my mouth had time to dissipate. I then walked over to a table where four portable breathalyzers were anxiously waiting to display the level of my inebriation.
Waiting for the timer to expire after finishing my first drink felt like an eternity. Then my phone beeped. I walked over to the counter, powered on the Breathometer Breeze, and launched the corresponding application on my iPhone. With the Breeze and my phone connected through Bluetooth, a process that took no effort on my part, I took a deep breath and exhaled into the Breeze for five seconds.
0.019%.
Next I picked up the Alcohoot. I plugged it into the headphone jack on my phone, launched its respective app, and waited for an indication that the device was ready. After another five seconds of blowing, the device clicked and my phone vibrated.
0.017%.
Then I powered on the BACtrack Mobile and waited for my phone’s Bluetooth connection to find the device. I tapped a button indicating that I would like to take a test. It took the device a few seconds to warm up, and as it initialized the app asked me to guess what I thought my BAC was. (I declined to answer—with two real tests behind me, it felt like cheating.) Once the device was ready, I blew into it until I heard the device click.
0.012%.
On to the last test. The BACtrack Vio is roughly the size of a pack of gum and designed to be hung on a keychain. It had a similar startup process to the BACtrack Mobile; once complete, I raised the device to my mouth and blew until the countdown timer expired.
0.031%.
Blowing into the each of the devices was no different than blowing bubbles in a glass of milk through a straw, save for the Vio, which offered a bit more resistance, akin to blowing up a balloon that just won’t stretch.
The night continued. I completed five tests and four drinks in as many hours. (It’s a tough job, but…) By the end of my test the Breeze placed my BAC at 0.082%. The Alcohoot had it at 0.114%. The BACtrack Mobile read 0.091%, and the smaller Vio displayed 0.117%. The final readings were all over the place, but all of them told me something I already knew: I was legally drunk.
I tested the devices on several occasions after that and began to detect a pattern. The Breeze consistently displayed the lowest BAC until it quit working altogether, displaying 0.000 for every test. (The company says my unit was defective.) The BACtrack Vio regularly returned the highest BAC reading. The BACtrack Mobile and Alcohoot were the most consistent devices and nearly all of their readings were within ± 0.008% of each other.
You’re probably wondering why Fortune would test smartphone breathalyzers. It’s simple: During the holidays there is a barrage of office parties, celebratory nights out, and casual drinking. And, save for the most citified readers, typically a drive home that follows. Today, we have mobile technology that allows us to quantify our inebriation and answer a very serious question: “Am I sober enough to drive?”
What I discovered while testing these breathalyzers runs quite the opposite of that proposition. BACtrack, Alcohoot, and Breathometer each shy away from encouraging users to trust their devices to make what ultimately could be a life or death decision. Rightfully so, as is evident from the test results. Each breathalyzer may indicate whether someone has alcohol in his or her system, but they’re not nearly trustworthy enough for such a decision.
How accurate are these breathalyzers relative to the ones used by law enforcement, you ask? Unfortunately, I was unable put the accuracy claims made by each company up against the machines used by professionals. (It’s against the internal policies of my local police department and sheriff’s office.)
To each company’s credit, each breathalyzer mobile app features the ability to call a cab and gives a time-based estimate of when you’ll be sober. Some of them also provide directions to the nearest hotel.
But that’s not the primary reason you download an app like this. In the United States the legal driving limit is a BAC of 0.08%. After my second drink on this particular night, I felt as if I was already incapable of safely operating a motor vehicle. My speech was beginning to slur. My coordination was less than stellar—I felt as if I could trip over my feet when walking. Yet according to the devices at my disposal, my BAC was somewhere between 0.036% and 0.057%. In theory, I could have sat behind the wheel of my car and legally driven across town. A scary thought, no doubt.
The next morning I woke up roughly 30 minutes after the time I was supposed to be sober. My head was throbbing. I grabbed a bottle of Advil and made a cup of coffee. As I sat at the dining room table looking over the chart of BAC results from the night before, I swore off ever testing breathalyzers again.
“Logged In” is Fortune’s personal technology column, written by Jason Cipriani. Read it on Fortune.com each Tuesday.
