01 / THE PROMISE
Why did Theranos fail?
Theranos promoted a broad blood-testing system, but its Edison device had limited capability and staff reported repeated quality-control failures. The court record says only twelve tests ever ran on the Edison. Other machines did much of the work.[1]
While Theranos sold its broad testing system, staff raised concerns about the accuracy of its own device. The checks meant to establish that its results were reliable kept failing.[1]
The machine needed to turn a small blood sample into a measurable signal, then verify that the signal gave a reliable result.
WATCH THE EXPLANATION
The test and the checks that failed.
Follow the sample through the assay, then see where the evidence behind the result broke down.
02 / THE MEASUREMENT
From blood to an answer.
The documented Theranos antibody test shows how the system was supposed to convert a sample into a measurable signal.
In 2015, the FDA cleared a Theranos test for antibodies to herpes simplex virus type 1. Its listed sample types included blood collected from a finger prick, so small-sample testing was possible for this assay.[2]
Matching antibodies attach to a coated surface.
Unattached material is removed.
Reagents produce a measurable light signal.
For this assay, the sample meets a surface coated with a protein from the virus. Matching antibodies bind to that surface; washing removes unattached material. Another reagent binds to the captured antibodies. After another wash, a chemical reaction produces light. The system measures that light and uses a calibration function to interpret the signal.[2]

The instrument also runs positive and negative controls. Those must fall within specified limits for the run to count as valid. Producing a signal is only one part of producing a trustworthy answer.[2]
One cleared test was not validation of the whole Theranos promise.
THE CRITICAL DISTINCTION
03 / THE FAILURE CHAIN
Where the checks broke down.
A former Theranos lab director described one problem at trial: damaged red blood cells. When those cells break open, their contents enter the sample and can interfere with testing. He testified that this happened more often in Theranos fingerstick samples than in its standard vein samples.[1]
Damaged cells do not explain every bad result. A smaller sample still needs protection from damage, and the test still needs verification.
Theranos used the Edison to test patient samples; the later miniLab was a separate development project. Commercial machines, some modified, performed other tests. The equipment doing the work differed from the machine the company promoted.[1]
Another lab director testified that proficiency testing used third-party machines while patient samples were run on Theranos devices. When results from the two were compared, the differences raised concerns about accuracy.[1]
A check on one machine cannot, by itself, establish the performance of another.
04 / THE OUTCOME
The results could not stand.
Federal inspectors examined Theranos in late 2015. Their findings identified quality-control failures involving both the Edison and commercial equipment. The laboratory deficiencies posed immediate jeopardy to patient health.[1] [3]
In its response to regulators, Theranos described repeated shifts in Edison control results and an unstable testing process. It acknowledged a long-term failure of the quality-control program and possible patient impact for every test reported from those instruments.[3]
Theranos said it did not know what fraction of results was affected or how large the errors were. A voided result could no longer be relied on, even though voiding it did not prove that its number was wrong.[3]
Theranos voided all patient test results reported from its Edison instruments. In July 2016, regulators issued sanctions, including revocation of the Newark laboratory’s certification.[3]
Elizabeth Holmes was also tried for fraud. She was convicted on four investor-related counts and acquitted on the patient-related counts decided by the jury.[4]
Theranos promised a smaller sample and a simpler machine, but had not established that its results were reliable enough for patients to trust.
THE SOURCE RECORD
Follow the evidence.
Primary documents for this companion to our Theranos episode:
- United States v. Holmes
Ninth Circuit · February 24, 2025
Device capability and controls: pp. 9–12. Trial testimony: pp. 24–26. - FDA clearance and assay summary, K143236
One HSV-1 antibody test · 2015
Indications and assay mechanism: PDF pp. 1, 3 and 5. - CMS sanctions letter to Theranos
July 7, 2016 · Scan of the original agency letter
Findings, retrospective controls and voided results: pp. 1, 7–8; sanctions: pp. 31–33. - Department of Justice verdict announcement
January 4, 2022 · Investor convictions and patient-count acquittals.
Court testimony is attributed where used. No single sample problem is presented as the explanation for every inaccurate result. Laboratory imagery is illustrative; diagrams explain the documented assay.
