Written and maintained by CASRAI Editorial Board
Last updated
Almost everyone meets this technique through a terminology problem. RT-PCR and qPCR are not the same thing, they are not abbreviations of each other, and the “RT” does not stand for “real time”. Getting that straight is the first useful thing this guide can do:
| Term | What it means |
|---|---|
| PCR | Polymerase chain reaction. Amplifies DNA. Result read at the end, usually on a gel. |
| qPCR | Quantitative PCR, also called real-time PCR. Amplification is measured as it happens, via fluorescence, so you can quantify the starting amount. |
| RT-PCR | Reverse transcription PCR. RNA is first converted to cDNA, then amplified. “RT” = reverse transcription. |
| RT-qPCR | Both at once: reverse-transcribe RNA to cDNA, then quantify in real time. This is what most people mean when they say “RT-PCR” for gene expression. |
So: qPCR quantifies. RT-PCR handles RNA. If you are measuring gene expression, you are almost certainly doing RT-qPCR. The rest of this guide covers how to run it, what makes a result trustworthy, and how to troubleshoot it.
How qPCR actually works
Conventional PCR tells you whether a product appeared. qPCR tells you when it appeared, and that timing is the measurement.
A fluorescent signal proportional to accumulated product is read at every cycle. Early cycles produce signal below the detection threshold; once amplification becomes measurable, signal rises exponentially, then plateaus as reagents deplete. The cycle at which fluorescence crosses a set threshold is the quantification cycle (Cq) — older literature and some instruments call it Ct or Cp.
The key relationship: more starting template means fewer cycles to reach the threshold, so a lower Cq. Cq is inversely related to starting quantity, and because amplification is exponential, each unit of Cq corresponds to a doubling — a difference of about 3.3 cycles corresponds to a tenfold difference in starting material at perfect efficiency.
Quantification uses only the exponential phase. The plateau is uninformative, which is exactly why endpoint PCR quantifies badly and why running an agarose gel at the end tells you much less than the curve does.
Chemistry: dye versus probe
| Intercalating dye (e.g. SYBR Green) | Hydrolysis probe (e.g. TaqMan) | |
|---|---|---|
| How it signals | Fluoresces when bound to any double-stranded DNA | A sequence-specific probe is cleaved during extension, separating reporter from quencher |
| Specificity | Non-specific — it reports primer-dimers and off-target product too | Specific — a third sequence must match for any signal |
| Melt curve possible | Yes, and you should always run one | No, and it is not needed |
| Multiplexing | Not practical | Yes, with differently-labelled probes |
| Cost and setup | Cheaper, faster to design | More expensive, more design work |
The practical rule: dye chemistry is fine when primers are well validated and you run melt curves; probe chemistry is worth the cost when specificity is critical, when you are multiplexing, or when the target sits in a sequence family where off-target amplification is likely.
The melt curve is not optional with dye chemistry
Because the dye reports any double-stranded DNA, a beautiful amplification curve can be primer-dimer. After cycling, the instrument ramps temperature and watches fluorescence fall as product denatures. A single sharp peak means one product. Two peaks, or a broad low-temperature shoulder, means you are quantifying something you did not intend. Skipping the melt curve is the most common way people report confident, meaningless dye-based qPCR data. Reading the peaks correctly is a separate skill: which shapes mean primer dimer, which mean a non-specific amplicon, and which confirmatory check settles the case are covered in our guide to melt curve analysis and primer dimers.
Quantification: absolute versus relative
Absolute quantification converts Cq to copy number using a standard curve built from a dilution series of known concentration. Necessary when the actual number matters — viral load, copy-number determination, GMO content.
Relative quantification compares expression between conditions, normalised to reference genes. This is what most gene-expression work does, usually via the comparative method (ΔΔCq), which assumes near-100% and closely-matched amplification efficiency between target and reference. Where that assumption does not hold, an efficiency-corrected method should be used instead — and you only know whether it holds if you have measured efficiency.
The standard curve, and what it tells you
Run a dilution series of your target and plot Cq against log input. The commonly used acceptance criteria in the field are an efficiency of roughly 90-110%, corresponding to a slope near -3.32, with a linearity (R²) above about 0.98.
Efficiency below range usually means inhibitors, poor primers, or degraded reagents. Efficiency above 110% is not a bonus — it generally indicates inhibition at high concentrations or pipetting error in the dilution series, and should be investigated rather than celebrated.
Reference genes: the most common serious error
Relative quantification is only as good as its normaliser, and this is where a great deal of published qPCR quietly fails.
GAPDH, ACTB and 18S rRNA are used by default and are frequently the wrong choice, because their expression is not actually stable across many treatments, tissues and disease states — the assumption that a “housekeeping gene” is constant is an assumption, not a fact, and it is routinely violated. If your reference gene shifts with your treatment, your normalised result reports that shift as though it were your target’s biology.
Good practice, and increasingly a reviewer expectation:
- Validate reference genes in your specific experimental system, rather than inheriting them from a previous paper.
- Use more than one. Multiple validated references, geometrically averaged, are far more robust than any single gene.
- Use a stability-ranking approach (algorithms such as geNorm, NormFinder and BestKeeper exist for exactly this) rather than eyeballing it.
- Report which references you used, why, and how you validated them.
Controls that determine whether the result means anything
| Control | What it catches |
|---|---|
| No-template control (NTC) | Contamination and primer-dimer. Any amplification here is a red flag. |
| No-RT control (-RT) | Genomic DNA contamination. Essential for RT-qPCR — without it you cannot tell cDNA amplification from gDNA amplification. |
| Positive control | A false negative caused by failed reagents, cycling, or an uncalibrated instrument (see thermal cycler validation). |
| Inter-plate calibrator | Plate-to-plate variation when an experiment spans multiple runs. |
The no-RT control deserves emphasis because it is the one most often skipped and the one that most often invalidates a result. Many primer pairs will happily amplify genomic DNA. If your DNase treatment was incomplete and you have no -RT control, you have no way to know that some or all of your “expression” signal is genomic.
MIQE: the reporting standard
The MIQE guidelines — Minimum Information for Publication of Quantitative Real-Time PCR Experiments (Bustin et al., Clinical Chemistry, 2009) — were written precisely because qPCR papers were being published without enough detail to evaluate or reproduce them. They set out what a paper should report: RNA quality and quantification method, reverse-transcription conditions, primer and probe sequences, amplification efficiency, Cq determination method, reference-gene selection and validation, controls, and the analysis method used. A 2025 revision, MIQE 2.0, replaced the original essential/desirable tiering with a single unified yes/no checklist spanning reagent preparation, sample preparation, reverse transcription, the qPCR protocol itself, and data analysis — cite MIQE 2.0 rather than the 2009 original in new methods sections.
Two reasons this matters beyond box-ticking. First, several journals now expect MIQE-consistent reporting, so it affects acceptance. Second, and more usefully, the checklist is a good experimental design tool — if you cannot fill it in, that is usually because something in the experiment was not controlled, and it is far cheaper to discover that before running the plates than at review.
Troubleshooting by symptom
Guessing at causes wastes plates. For each symptom below, run the single isolating experiment first — it tells you which candidate cause you actually have before you start changing reagents, primers, or protocol.
No amplification (flat line, no Cq)
Either the reaction itself failed — degraded enzyme, degraded or omitted primers/probe, a wrong or corrupted cycling program, a reagent left out of the master mix — or there is effectively no amplifiable template, which happens when extraction or reverse transcription failed outright rather than merely performed poorly.
Isolating experiment: spike a small volume of a known-good positive control (a plasmid standard, a previously-amplifying cDNA, or an assay-specific positive control) into an aliquot of the same master mix and run it in the same plate as the failed sample. If the positive control also fails to amplify, the fault is in the reaction itself — reagents, program, or instrument — not the sample. If the positive control amplifies normally while the sample stays flat, the fault is upstream: quantify the sample’s nucleic acid directly (Qubit or a UV-Vis reading) to confirm template is actually present, and confirm the reverse-transcription step was actually run if the input was RNA.
Late amplification (Cq higher than expected)
Too little template, degraded RNA or cDNA, inhibitors carried over from RNA extraction (phenol, ethanol, guanidine salts, heparin in blood samples), a low-efficiency primer set, or a reverse-transcription step that ran but ran poorly.
Isolating experiment: run a two- or three-point dilution series of the template (e.g. neat, 1:5, 1:25) alongside the original reaction. Genuinely low template gives a Cq shift that tracks the dilution factor — about 3.3 cycles per tenfold step. Inhibition instead shows a Cq that improves by more than the dilution predicts, because the inhibitor dilutes out along with the target. The more decisive version of this test spikes a fixed amount of an unrelated exogenous target (an internal positive control) into both the sample and an equal volume of nuclease-free water; a delayed Cq for the spiked target in the sample relative to water isolates inhibition directly, independent of how much real target is present.
Amplification in the no-template control (NTC signal)
Contamination of a shared reagent or primer-dimer — not sample carryover, since the NTC by definition never saw a sample. Check the NTC melt curve first: a low-temperature peak points to primer-dimer, a peak matching your product’s Tm points to contamination. Published measurements of this specific failure mode found the two artefact classes correlate with neither Cq nor calculated efficiency, so a “close enough” Cq is not evidence the NTC is clean — the melt curve is the actual diagnostic.
Isolating experiment: rerun the NTC using a freshly-thawed reagent aliquot and freshly-aliquoted water, prepared in the pre-PCR area with a new set of filter tips. If the signal disappears, the original reagent stock or workspace was contaminated — remake reagents, use filter tips throughout, and physically separate pre- and post-PCR work. If the signal persists with everything fresh, the primers themselves are dimerising; redesign them or raise the annealing temperature.
High replicate variability (high CV between technical replicates)
Usually pipetting error at small volumes, incomplete mixing of the master mix, or a plate-position (edge) effect. Variability that appears only at high Cq is expected on its own — near the detection limit, stochastic sampling of a handful of template molecules genuinely varies run to run, and that is not a technique problem to chase.
Isolating experiment: to separate pipetting error from mixing or plate effects, dispense the same already-mixed master-mix aliquot into eight to ten wells spread across the plate — center and edge positions both — and compare the Cq spread against a second set of eight to ten wells each pipetted independently from the stock reagents. A high CV that shows up only in the independently-pipetted set isolates pipetting technique as the cause; a high CV that shows up even in the single shared aliquot, correlated with well position, isolates a plate or thermal edge effect instead. Prepare a master mix for real experiments either way, and avoid dispensing volumes below about 2 µL, where pipette accuracy degrades sharply.
Multiple melt peaks
Non-specific amplification or dimer. Raise the annealing temperature, redesign primers, or move to probe chemistry. See melt curve analysis and primer dimers for reading the peak shapes themselves.
Efficiency outside 90-110%
Below range: inhibitors, suboptimal primers, degraded reagents. Above range: almost always a dilution-series or pipetting artefact in the standard curve rather than genuinely super-efficient amplification — the inhibition-isolating spike-in test above applies here too.
Signal in the no-RT control
Genomic DNA contamination. Treat with DNase, or design primers that span an exon-exon junction so genomic template cannot amplify.
What research administrators and lab managers should know
- Genomics cores usually run qPCR and digital PCR on a recharge basis, including assay design and validation. Budget instrument time and per-plate consumables explicitly on grant applications rather than absorbing them.
- Reagent cost scales badly with replication. Proper technical and biological replication, plus the controls above, is where the budget actually goes — and cutting replicates to save money is what makes the data unpublishable.
- Data deposition and reporting. Journals increasingly ask for raw amplification data and MIQE-consistent detail. Retaining instrument run files matters, and the data management plan should say where they live.
- Clinical or diagnostic use is a different regulatory world — CLIA, validation requirements and quality systems apply. A research-use assay does not transfer to clinical use without that work.
Frequently asked questions
What is the difference between RT-PCR and qPCR?
RT-PCR uses reverse transcription to convert RNA into cDNA before amplification. qPCR quantifies amplification in real time. They address different problems and are frequently combined as RT-qPCR, which is what most gene-expression work uses.
Does RT stand for real time?
No. In RT-PCR, RT means reverse transcription. Real-time PCR is qPCR. The overlap in abbreviations is the source of most of the confusion around this technique.
What is a Cq or Ct value?
The cycle number at which fluorescence crosses a defined threshold. Lower Cq means more starting template. Cq is the current preferred term; Ct and Cp mean the same thing.
What is a good qPCR efficiency?
Roughly 90-110% is the commonly accepted range, with R² above about 0.98. Values above 110% usually indicate an artefact rather than better performance.
Do I really need a no-RT control?
For RT-qPCR, yes. It is the only way to demonstrate that your signal comes from cDNA rather than contaminating genomic DNA, and it is the control most often omitted from otherwise careful experiments.
Can I use GAPDH as my reference gene?
Only if you have validated that it is stable in your specific system. GAPDH, ACTB and 18S are widely used and frequently unstable across treatments and tissues. Validate, and use more than one reference.
Why do I need a melt curve?
With intercalating-dye chemistry, the dye reports any double-stranded DNA, including primer-dimer and off-target product. The melt curve is what tells you the signal you quantified came from a single intended product.
How do I tell PCR inhibition from a genuinely low-template sample?
Run a dilution series, or spike a fixed amount of an internal positive control into the sample and into an equal volume of water. A Cq improvement on dilution bigger than the dilution factor predicts, or a delayed spiked-control Cq in the sample versus water, isolates inhibition rather than true low quantity.
What does it mean if my no-template control amplifies?
Reagent contamination or primer-dimer — not sample carryover, since the NTC never held a sample. Check the NTC melt curve, then rerun it with a fresh reagent aliquot in the pre-PCR area: if the signal disappears, the previous stock or workspace was contaminated; if it persists, the primers themselves are dimerising.
