Ct Values Above 35: Should You Trust Them?

A Ct above 35 is not automatically junk — but it's guilty until proven innocent. The honest answer is that most Ct values between 35 and 40 represent either primer-dimer amplification, low-level contamination, or stochastic noise from single-digit template copies, not meaningful biological signal. Before you include a late Ct in your analysis, you need to confirm it with melt curve data, NTC behavior, and replicate consistency. If you can't do all three, drop it.

That said, there are legitimate situations where a Ct of 36 or 37 is real: a genuinely low-abundance transcript in a small cell population, a rare splice variant, a viral target near the limit of detection. The question isn't really "is 35 a magic cutoff?" — it's "do I have enough evidence that this amplification event reflects my actual target?" Let's walk through how to make that call.

Why Late Ct Values Are Unreliable by Default

The math of qPCR works against you at low template concentrations. At a Ct of 35 with 100% efficiency, you're starting from roughly 3–30 copies of template depending on your reaction setup. At that level, Poisson sampling dominates: some replicate wells get 5 copies, some get zero, some get 1. This is why replicate Ct values that are tight at Ct 20 (±0.1) blow apart at Ct 35 (±1.5 or worse).

There are also competing amplification events that become visible only when your real target is scarce:

• Primer-dimers: Most primer pairs will produce some dimer product if given 40 cycles. With SYBR Green or other intercalating dyes, dimers generate fluorescence indistinguishable from target amplification until you check the melt curve. Dimers typically melt 5–10°C below your target amplicon — look for a peak around 72–78°C when your target melts at 82–88°C.

• Genomic DNA carryover: If you're measuring a transcript and your primers don't span an intron (or the intron is small enough to amplify from gDNA), residual genomic DNA after DNase treatment can produce a late Ct. This is what your no-reverse-transcriptase (NRT) control is for.

• NTC contamination: Aerosol contamination of your no-template control is surprisingly common in busy labs. If your NTC shows a Ct of 37–39, you cannot trust a sample Ct that's within 5 cycles of it. Full stop. You need at least a 5-Ct gap — ideally more — between your NTC signal and your lowest sample Ct to claim that signal is from your template and not from contamination (Bustin et al., 2009, The MIQE Guidelines).

• Stochastic amplification artifacts: At very low copy numbers, random mispriming events can initiate exponential amplification of off-target products. Even TaqMan assays, which require probe hybridization, can occasionally show nonspecific late-cycle signal, though it's far less common than with SYBR-based chemistries.

The Practical Decision Framework

Here's how I evaluate a Ct above 35, in order:

1. Check your NTCs. If your NTC is negative (no amplification through cycle 40), that's a good start. If your NTC shows a Ct of 38 or 39, your sample Ct of 36 is suspect. On a CFX96 or QuantStudio, pull up the amplification curves for your NTCs alongside your samples. Look at the raw ΔRn curves, not just the reported Ct. Sometimes the software calls a "Ct" on what is clearly baseline drift, not real amplification.

2. Check the melt curve (SYBR) or multicomponent plot (TaqMan). For SYBR Green assays (PowerUp SYBR, Luna Universal, etc.), a single sharp melt peak matching your positive control confirms the right product. A broad peak, a shoulder, or a second peak at a lower Tm means primer-dimer or off-target amplification is contributing to — or entirely responsible for — that Ct value. For TaqMan assays, look at the multicomponent plot: is the reporter dye (FAM, etc.) genuinely separating from the baseline, or is it just noise that the algorithm fit a threshold line to?

3. Check replicate consistency. If you ran triplicates and got Ct values of 35.2, 36.8, and undetermined, that's not a reliable measurement. For Ct values above 35, I want to see all three replicates amplify and agree within ±1.0 Ct to even consider using the data. A standard deviation of 0.3–0.5 Ct is normal for mid-range targets; at Ct 35+, anything under 1.0 Ct SD across replicates is acceptable but should still be flagged. If one replicate drops out entirely, the most honest interpretation is often "at or below the limit of detection."

4. Consider your assay's validated dynamic range. If you ran a standard curve during assay validation (and you should have — MIQE essentially requires it), what was the lowest standard that still fell on the linear curve? If your last reliable standard was 10 copies at Ct 34, a sample Ct of 37 is extrapolating beyond your validated range. You can report it as "detected but below the limit of quantification (LOQ)," which is a meaningful and honest result, but you shouldn't calculate a fold change from it.

5. Run a gel or fragment analysis if the result matters. If a late Ct is going to change the conclusion of your experiment — gene X is expressed vs. not expressed in this tissue — run 5 µL of the qPCR product on a 2% agarose gel or a Bioanalyzer/TapeStation. A single band at the expected amplicon size (say, 120 bp) is strong evidence the Ct is real. Multiple bands or a smear near the primer front tells you the Ct is artifact.

When Late Ct Values Are Genuinely Useful

Not every Ct above 35 is noise. Some real-world scenarios where late Ct values carry biological meaning:

Pathogen detection. In clinical or environmental virology, the question is often binary: is the target present or not? A SARS-CoV-2 RT-qPCR with a Ct of 37 on the N gene might represent a patient at the tail end of infection with a very low viral load. In these contexts, labs typically use a validated cutoff (often Ct 37 or 38 for specific assays) and report anything above it as "not detected" or "inconclusive" rather than "negative."

Rare transcripts in small populations. If you're looking at FOXP3 expression in sorted regulatory T cells and you only got 500 cells, a Ct of 36 for a low-abundance target might be entirely real — you just don't have much RNA to work with. In these cases, pre-amplification (e.g., with Fluidigm-style targeted pre-amp) or switching to digital PCR gives you much better quantification at the low end.

Contamination monitoring. If you're using qPCR to test whether a cell line is mycoplasma-free, even a Ct of 38 on a mycoplasma-specific assay is a concern that warrants follow-up, precisely because the cost of ignoring it is high.

What About Setting a Hard Cutoff?

Many labs use Ct 35 as a blanket cutoff: anything above it gets called "undetermined." This is reasonable as a default — it's conservative and avoids most false positives. But it's arbitrary. The right cutoff depends on your assay, your chemistry, and your NTC behavior.

A better approach is to set your cutoff based on your NTC and your standard curve:

• If your NTCs are consistently negative, your cutoff can be higher (37–38).
• If your NTCs occasionally show signal at Ct 38–39 (common with SYBR assays for high-copy targets like 18S or GAPDH where template aerosol is everywhere), set your cutoff at least 5 Ct below the mean NTC Ct.
• If your lowest validated standard (e.g., 10 copies) amplifies at Ct 33, don't quantify anything above Ct 33 — report it as detected/not detected.

The MIQE guidelines (Bustin et al., 2009) recommend against using a one-size-fits-all cutoff and instead advocate for assay-specific LOD and LOQ determination. In practice, most academic labs don't do formal LOD/LOQ studies for every primer pair, so Ct 35 persists as a pragmatic line in the sand. Just know it's a convention, not a physical law.

How to Report Late Ct Data

If you decide a late Ct is real and you want to include it in a ΔΔCt or Pfaffl calculation, be transparent:

• Report the actual Ct values in your supplementary data, not just the fold changes.
• Note which samples had Ct values above 35.
• If replicates were inconsistent, say so.
• Consider reporting the result as "detected at low level" rather than assigning a precise fold change. A ΔCt-derived fold change calculated from Ct 36.5 ± 1.2 has enormous error bars that will overlap with "no expression" in most experiments.

For statistical analysis, remember: run your t-tests or ANOVA on ΔCt values, not on fold changes. Fold changes are exponentially transformed and violate the normality assumptions of parametric tests. If some samples have undetermined Ct values, you need to make a decision — assign them the maximum cycle number (40), exclude them, or use a censored data approach. Each choice has implications, and you should state which one you used.

The Quick Version

Trust a Ct above 35 only if: your NTC is clean, your melt curve shows a single correct peak, your replicates agree within ±1.0 Ct, and the value falls within your assay's validated range. If any of those checks fail, treat the result as "at or below the detection limit." When you're analyzing a dataset with a mix of reliable and borderline Ct values, VoilaPCR flags late Ct values and NTC issues automatically so you can make these judgment calls without manually auditing every well.