It is a scenario that unsettles a lot of breeders the first time it happens. A sample drawn Monday morning at one clinic reads 4.2 ng/mL. The same bitch, sampled the same day and sent to a reference laboratory, comes back at 7.8. One of them must be wrong.
Usually neither is wrong. Understanding why progesterone results differ across platforms turns a confusing result into a manageable one — the measurement is genuinely more variable than most people expect.
- Different assay platforms produce genuinely different numbers on the same canine sample, by margins large enough to move a breeding decision.
- Because timing depends on detecting a rise, internal consistency matters more than absolute accuracy. Pick one platform and stay on it for the whole cycle.
- Check the unit on every report. ng/mL and nmol/L differ by a factor of about 3.18.
- Separate serum promptly and avoid hemolysis. Handling changes the answer.
Why progesterone results differ between methods
Several distinct technologies are in routine use for canine progesterone, and they do not produce interchangeable numbers.
- Radioimmunoassay (RIA). Historically the reference method, and the basis for many of the threshold values breeders still quote. Increasingly rare in practice.
- Chemiluminescence immunoassay (CLIA and ECLIA). The current workhorse of reference laboratories, adapted from human clinical chemistry platforms.
- Dry-slide and cartridge-based immunoassays. In-house bench analyzers offering same-day results in the clinic.
- Fluorescence immunoassay, including lateral-flow point-of-care systems. Widely used for in-clinic veterinary testing.
- Semi-quantitative visual or reader-based kits. Report bands or ranges rather than a precise concentration.
The gaps between them have been measured directly. Nöthling and De Cramer compared the Siemens Immulite chemiluminescence assay against the Coat-A-Count radioimmunoassay on blood collected at the same time from the same bitches. On average, the Immulite result was just 85% of the RIA value, with a wide scatter around that mean (95% confidence interval 58% to 112%), and 88% of Immulite readings were lower than their RIA counterparts.
Read that confidence interval again, because it is the number that matters. A 15% average shift you could in principle correct for. A spread running from 42% low to 12% high on an individual sample you cannot.
In-house analyzers show the same pattern. Østergård Jensen and colleagues, publishing in Veterinary Clinical Pathology, validated a dry-slide immunoassay for canine plasma against both chemiluminescence and mass spectrometry. They found a significant difference between the dry-slide and chemiluminescence results, and a proportional bias when the dry-slide method was compared with mass spectrometry — with chemiluminescence correlating more closely to the mass spectrometry reference.
Put plainly: a 2.0 ng/mL threshold established on one platform is not automatically a 2.0 ng/mL threshold on another. This has direct consequences for how you read your own analyzer’s chart, which is covered in more depth in why your analyzer’s chart doesn’t match the textbook.
The one rule that solves most of this
Pick one testing method and stay on it for the entire cycle. Internal consistency matters more than absolute accuracy.
Because canine ovulation timing is fundamentally about detecting a rise rather than hitting an exact value, an assay that reads consistently low is still perfectly usable, as long as every sample in that cycle went through the same instrument. What breaks the reading is mixing platforms mid-cycle, because you can no longer tell whether a change reflects the dog or the machine.
If you genuinely must switch mid-cycle, note it prominently in your records and expect a step change that has nothing to do with physiology.
Units: ng/mL versus nmol/L
This one catches people out constantly, particularly when a report comes from a laboratory outside the United States or from a platform configured for human medicine. Progesterone is reported in nanograms per millilitre (ng/mL) in most North American veterinary practice and in nanomoles per litre (nmol/L) in much of Europe and in many clinical chemistry systems.
| ng/mL | nmol/L (approx.) | Interpretation |
|---|---|---|
| 1.0 | 3.2 | Baseline |
| 2.0 | 6.4 | LH surge / Day 0 |
| 2.5 | 8.0 | Day 0 confirmed |
| 5.0 | 15.9 | Ovulation |
| 10.0 | 31.8 | Approaching fertile window |
| 20.0 | 63.6 | Established luteal phase |
The trap
A result of “6.4” looks like a comfortable post-ovulation number to anyone reading in ng/mL. In nmol/L it means the bitch has only just surged and the breeding is still four days out. The arithmetic is easy — multiply ng/mL by 3.18, or divide nmol/L by 3.18. The danger is failing to notice the unit at all.
Sample handling changes the answer
Progesterone is reasonably robust compared with some hormones, but it is not indestructible. The Østergård Jensen validation put numbers on this, and they are larger than most breeders assume:
| Handling condition | Measured effect |
|---|---|
| Stored at room temperature for 12 to 24 hours | Concentrations fell to 57% to 96% of the initial value |
| Frozen at −80°C | Concentrations reduced by 17% to 27% |
| Hemolytic and lipemic samples | Minor changes observed |
The freezing result is the counterintuitive one. Freezing feels like preservation; here it was associated with a measurable loss rather than a hold. That alone is a reason not to bank samples and run them later during a cycle you are actively timing.
- Separate serum promptly. Leaving whole blood sitting on the clot allows drift.
- Avoid hemolysis and lipemia. Both can interfere with immunoassay readings. A fasted sample and clean venipuncture help.
- Confirm the accepted sample type. Some analyzers are validated for serum only, others accept whole blood or plasma, and some manufacturers supply separate kits for each. EDTA and heparin are not always interchangeable.
- Do not assume freezing preserves the value. See the table above.
- Sample at a consistent time of day. This does not eliminate variation, but it removes one source of noise from a serial curve.
Point-of-care systems differ in what they accept. The PETlife Pro-DX, for example, supplies the canine progesterone (cProg) test in separate whole-blood and serum/plasma versions, both running on 75 µL with results in about 15 minutes. Using the kit that matches your sampling workflow removes one avoidable source of variation, and testing on site removes the room-temperature transit window entirely.
When semi-quantitative kits are enough, and when they are not
Kits that report a range rather than a number have a legitimate role. If your question is “has the rise started yet, or should I test again in three days,” a categorical answer is entirely sufficient and considerably cheaper.
Where they fall short is at the decision point. De Cramer is blunt about it: progesterone test kits performing semiquantitative evaluation proved much less useful and reliable than quantitative assays that give an accurate numerical value. Distinguishing 2.2 from 3.4 ng/mL can move Day 0 by a full day, and with frozen semen a full day is the whole margin. For high-stakes breedings, use a quantitative method through the critical window even if a screening kit got you there.
What to record alongside the number
Keeping the platform recorded alongside each result is what makes a multi-year record genuinely comparable. Without it, a value from two seasons ago is a number you cannot safely trust against today’s. The free Pup Planner app keeps test results plotted against the individual dog, which makes the series readable rather than a stack of loose slips.
Frequently asked questions
Which progesterone test is the most accurate?
Chemiluminescence assays correlate most closely with mass spectrometry among the widely available methods, as Østergård Jensen and colleagues found, and radioimmunoassay was the historical reference. But for ovulation timing, the practical answer is that consistency across the cycle matters more than which platform you chose.
How do I convert nmol/L to ng/mL?
Divide by 3.18. So 6.4 nmol/L is about 2.0 ng/mL, and 15.9 nmol/L is about 5.0 ng/mL. To go the other way, multiply ng/mL by 3.18.
Can I switch labs partway through a cycle?
It is best avoided. If circumstances force it, record the switch clearly and expect a step change in the numbers that reflects the instrument rather than the dog. Do not read that step as a physiological event.
Sources
Nöthling JO, De Cramer KGM. Comparison of progesterone assay by chemiluminescence or radioimmunoassay for clinical decision-making in canine reproduction.
Østergård Jensen S, et al. Validation of a dry-slide immunoassay for progesterone analysis in canine plasma in a clinical setting. Vet Clin Pathol. 2022.
De Cramer K. Simplifying the Canine Reproductive Cycle Using the Concentration of Progesterone. IDEXX Laboratories, CLD-12211-00.
Keep reading
This article is educational and is not a substitute for veterinary advice. Assay selection, sample handling protocols, and result interpretation should be discussed with your veterinarian or testing laboratory.