Reading a Melt Curve, and Getting Efficiency From a Standard Curve
8 min read · Updated November 11, 2026
Two quality checks stand between a qPCR run and a fold change you can defend: a melt curve that shows one product, and a standard curve whose slope gives an efficiency near 2. Both are usually glanced at rather than read.
They answer different questions. The melt curve asks whether one thing was made; the standard curve asks how well it was made across a range of inputs. An assay can pass one and fail the other.
What a melt curve is measuring
After the last cycle the instrument heats the plate slowly and watches fluorescence. As the product denatures, the intercalating dye is released and the signal falls; the negative first derivative of that curve peaks at the temperature where the most product melts, which is what the software plots.
That peak is a property of the product's length and GC content, not of your primers. A single sharp peak at a temperature consistent between wells means one dominant product is present. It does not mean the product is the right one — a non-specific amplicon of similar composition melts at a similar temperature — but two peaks is unambiguous evidence that at least two products are there.
The failure modes, in the order they occur
- Two distinct peaks: two products. Usually primer-dimer plus amplicon, or a mispriming product. The lower-temperature peak is often the dimer, because short products melt first.
- One peak with a shoulder on the low-temperature side: usually dimer that melts close to the amplicon, and worth treating as two products. Diluting the primers and raising the annealing temperature separates them.
- One peak at the wrong temperature for the amplicon: the product is not what you designed. Run it on a gel before believing any Ct from that well.
- No peak, with a late Ct: primer-dimer only. The well amplified something, and it was not the target.
- A peak in the no-template control: the primers form product without template, and every Ct in the run is contaminated by it.
Worked: slope to efficiency
A standard curve plots Ct against log10 of the input, usually over five 10-fold dilutions. The slope of that line is negative, and efficiency comes out as E = 10^(−1/slope), where E is the fold amplification per cycle. A slope of −3.32 gives 2.0008, which is 100.1% — the value a perfect doubling produces, and the reason −3.32 is the number everyone quotes.
The useful part is how fast the number degrades. A slope of −3.59 gives 1.8991, or 89.9%; −3.5 gives 1.9307, or 93.1%; −3.1 gives 2.1017, or 110.2%; and −3.9 gives 1.8047, or 80.5%. Ten percentage points of efficiency — 100% down to 90% — is a slope error of 0.27, and a scatter that size is ordinary across five points of a hand-made dilution series, so an efficiency quoted to three decimal places from them is over-claimed. The reverse is the reassuring half: a 0.08 slope error moves the efficiency only to 103.5%, which is why published windows are as wide as 90–110%.
Enter efficiency as the per-cycle fold, not as a percentage. 90% efficient is 1.90. Entering 95 for a 95%-efficient primer pair is the single commonest way a Pfaffl calculation goes wrong, and it goes wrong silently because the arithmetic still produces a number.
Why the efficiency is not allowed to be optional
2^−ΔΔCt assumes both amplicons double every cycle. When they do not, the Pfaffl ratio replaces it, and the size of the correction is worth seeing once. With the target at 1.90 and the reference at 2.00 over a three-cycle difference, the Pfaffl ratio is 6.86 against 8.00 from the shortcut — a 17% overstatement on the same Ct values.
The calculator prints the Livak answer beside the Pfaffl one for exactly this reason: the difference between the two is the size of the correction, and on a well-behaved assay it is small enough to ignore knowingly rather than unknowingly.
What the two checks still do not tell you
- Whether the product is the intended amplicon. A melt curve distinguishes products by their melting behaviour, not by their sequence.
- Whether the template is contaminated with genomic DNA, unless you ran a no-reverse-transcriptase control. A melt peak at the amplicon's own temperature cannot tell cDNA from gDNA.
- Whether the efficiency measured on standards applies to your samples. Inhibitors in a tissue extract shift the curve without changing the slope of a clean dilution series.
- Whether the reference gene is stable in your conditions. That is measured across your samples, not on a standard curve.
FAQ
What does a melt curve with two peaks mean?
That at least two products are present, which almost always means the Ct in that well is a mixture and cannot be interpreted as target abundance. Primer-dimer plus amplicon is the usual pair, with the dimer melting lower because it is shorter.
How do I calculate qPCR efficiency from a standard curve?
E = 10^(−1/slope), where the slope comes from plotting Ct against log10 of the input. A slope of −3.32 gives 2.00, or 100% efficiency; −3.59 gives 1.90, or 90%. Enter the result as a per-cycle fold, not as a percentage.
Is 90% efficiency good enough for qPCR?
Usually yes, if both amplicons are measured and the ΔΔCt shortcut is not applied blindly. 90–110% is the range most labs accept, and outside it the assay should be redesigned rather than corrected for.
Can a single melt peak still be the wrong product?
Yes. The melt curve reports melting behaviour, not sequence. A single peak at the expected temperature is consistent with the right amplicon and also with a similar-length non-specific product, which is why a gel or a sequencing check accompanies a new assay.
Related references
Related tools
Fold change from Ct values, with propagated error and efficiency-matched method.
Ranked PCR primer pairs from a template, with Tm, GC and dimer checks.
Primer Tm, GC content and molecular weight.
Product size and position predicted from a template and two primers.