PCR Cycling Conditions Reference
A standard PCR program repeats three temperature steps — denaturation, annealing and extension — for 25–35 cycles, bracketed by an initial denaturation and a final extension. This reference gives generic starting temperatures and times for a 3-step protocol and explains how to set the annealing temperature and extension time for your reaction.
Standard 3-step PCR program
| Step | Temperature | Time | Notes |
|---|---|---|---|
| Initial denaturation | 94–98 °C | 30 s – 3 min | Once, before cycling; fully melts template (higher/shorter for high-fidelity enzymes) |
| Denaturation | 94–98 °C | 10–30 s | Per cycle; melt the new product |
| Annealing | 45–72 °C | 15–30 s | Set from the lower primer Tm, by polymerase: ~3–5 °C below it for Taq, at or above it for Q5 and Phusion |
| Extension | 68–72 °C | by amplicon length | 72 °C for Taq; ~68–72 °C for proofreading enzymes |
| Final extension | 72 °C | 5–10 min | Once, after cycling; completes products / adds A-overhangs |
| Hold | 4–10 °C | ∞ | Storage until retrieval |
| Cycles | — | 25–35 | More cycles for low-template; too many increases artifacts |
These are general starting points across polymerase families, not vendor-specific settings. Always follow your polymerase's datasheet for exact temperatures, times and buffer.
The three steps
Each cycle has three steps. Denaturation at 94–98 °C melts the double-stranded template (and, after cycle one, the new product) into single strands. Annealing, typically somewhere between 45 and 72 °C, lets the primers bind their complementary sites. Extension at 68–72 °C is where the polymerase synthesizes the new strand. These three steps repeat 25–35×, bracketed by a one-off initial denaturation before the first cycle and a one-off final extension after the last.
Setting the annealing temperature
Work from the lower of the two primer Tm values — but how far from it depends on the polymerase, and on which Tm you calculated. For Taq-type enzymes the usual starting point is 3–5 °C below the lower primer Tm; NEB's own calculator uses Tm − 5 °C for its Taq buffers. High-fidelity enzymes go the other way. NEB anneals Q5 at Tm + 1 °C and Phusion at 0.93 × Tm + 7.5 °C (about Tm + 3 °C for typical primers), each on the Tm its calculator gives in that enzyme's own high-salt buffer. For Q5 that Tm runs several degrees above what a generic calculator reports for the same primer, which is why NEB tells anyone using a Tm from elsewhere to anneal Q5 at no less than Tm + 3 °C.
An offset only means something against the Tm it was defined on, so take the rule and the Tm from the same place — your polymerase's supplier. If the annealing temperature is too high, the primers do not bind and you get no product; if it is too low, they bind non-specifically and you get extra, non-specific bands. When you are unsure of the optimum, run a gradient PCR across a temperature range, or use a touchdown program (below) to converge on the correct product automatically.
Setting the extension time
Scale the extension time to the amplicon length and the polymerase family. Standard Taq is comparatively slow; proofreading / high-fidelity enzymes are faster, so a long template that needs minutes with Taq may need far less time with a high-fidelity enzyme. Use these rules of thumb as a starting point and confirm against your enzyme's datasheet.
| Polymerase family | Extension rate |
|---|---|
| Standard Taq | ~60 s per kb |
| Proofreading / high-fidelity | ~15–30 s per kb (faster; check your enzyme) |
Two-step and touchdown PCR
When the primers' annealing Tm is high — roughly ≥ 68 °C — the annealing and extension steps can be combined into a single step at 72 °C. This two-step PCR simplifies the program because the primers anneal efficiently at the extension temperature. Touchdown PCR instead starts the annealing temperature high and steps it down over the first cycles: the early high-stringency cycles favor the correct product so it dominates before any non-specific products can accumulate.
Frequently asked questions
What annealing temperature should I use?
It depends on the polymerase, and on which Tm you calculated. For Taq-type enzymes, start about 3–5 °C below the lower of your two primers' Tm values. High-fidelity enzymes in high-salt buffers go the other way: NEB's calculator anneals Q5 at the lower Tm + 1 °C and Phusion at roughly Tm + 3 °C, on the Tm it computes in that enzyme's own buffer — and NEB advises annealing Q5 at no less than Tm + 3 °C if your Tm came from any other calculator. Take the rule and the Tm from the same supplier. Too high and the primers will not bind, giving no product; too low and they bind non-specifically, giving extra bands. A gradient PCR or touchdown program helps find the best temperature.
How long should the extension step be?
Scale the extension time to the amplicon length and the polymerase family. As a rule of thumb, standard Taq extends at about 60 s per kb, while proofreading / high-fidelity enzymes are faster at roughly 15–30 s per kb — always check your enzyme's datasheet.
How many cycles should I run?
Most reactions use 25–35 cycles. Use more cycles when starting from very little template, but too many cycles increases non-specific products and artifacts as reagents are depleted.
What is touchdown PCR?
Touchdown PCR starts the annealing temperature a few degrees above the expected optimum and steps it down over the first several cycles. The early high-stringency cycles favor the correct product, which then dominates as the temperature drops — a simple way to improve specificity without optimizing a single annealing temperature.
Why are there separate initial denaturation and final extension steps?
The initial denaturation runs once before cycling to fully melt the template (and, for hot-start enzymes, to activate the polymerase). The final extension runs once after cycling to let the polymerase finish any incomplete products and, for Taq, add the 3' A-overhangs used in TA cloning.
See also
Sources
- 1PCR Protocol for Taq DNA Polymerase with Standard Taq Buffer (NEB #M0273)New England BiolabsThe Taq half of the PROGRAM table and the Standard Taq row of the EXTENSION table. Fetched and read in full. Backs: initial denaturation 95 °C 30 s (and the page's "30 s - 3 min" upper end via "a longer initial denaturation of 2-4 minutes at 95°C" for difficult templates, 5 min for colony PCR); per-cycle denaturation 95 °C 15-30 s; annealing 45-68 °C, 15-60 s; extension "1 minute/kb" (this is the page's "~60 s per kb" for standard Taq); final extension 5 minutes; Hold 4-10 °C; Cycles "25-35 cycles yield sufficient product... Up to 45 cycles may be required to detect low-copy-number targets" (the page's "More cycles for low-template"); the two-step section ("When primers with annealing temperatures above 65°C are used, a 2-step thermocycling protocol is possible"); and the FAQ claim about A-overhangs ("The PCR products generated using Taq DNA Polymerase contain dA overhangs at the 3'-end; therefore, the PCR products can be ligated to dT/dU-overhang vectors"). It also backs the gradient advice in the annealing section ("Annealing temperatures can be optimized by doing a temperature gradient PCR starting 5°C below the calculated Tm"). IMPORTANT DISAGREEMENT with the page: NEB's recommended extension temperature for Taq is 68 °C, not the 72 °C the page's Extension row asserts, and NEB's final extension is at 68 °C too. NEB returns 403 to curl and WebFetch; fetched in a real browser.
- 2PCR Protocol for Phusion High-Fidelity DNA Polymerase (NEB #M0530)New England BiolabsThe high-fidelity half of both tables, and the numbers match exactly. Fetched and read in full. Backs: the EXTENSION table's "Proofreading / high-fidelity ~15-30 s per kb" (Phusion's thermocycling table states 15-30 seconds/kb verbatim); the PROGRAM table's "Final extension 72 °C, 5-10 min" (Phusion states 5-10 minutes — NEB's Taq protocol says only 5 min, so the 10 min end comes from here); the 98 °C end of the "94-98 °C" denaturation rows (initial denaturation 98 °C 30 s, per-cycle 98 °C 5-10 s, which is also the source of the page's "10-30 s" per-cycle denaturation being achievable); annealing 45-72 °C for 10-30 s; 25-35 cycles; Hold 4-10 °C; and the existence of a two-step program with no separate annealing step. For a second high-fidelity data point, NEB Q5 (#M0491, https://www.neb.com/en-us/protocols/2013/12/13/pcr-using-q5-high-fidelity-dna-polymerase-m0491 — also fetched) gives 20-30 s/kb for gDNA and a 2-min final extension. Browser-only, as above.
- 3NEB Tm Calculator — Help (method, and why its Tm differs from other calculators)New England BiolabsThe high-fidelity half of 'Setting the annealing temperature' and the matching FAQ answer. Read 2026-09-25 in a browser (the page is a client-side app). Backs: that the calculator's Tm is buffer-specific ("designed to take into account the buffer conditions of the amplification reaction based on the selected NEB polymerase", SantaLucia 1998 with Owczarzy 2004 salt correction, Schildkraut for Phusion); and that Tm values from other calculators generally underestimate the Tm for Q5, with the advice to raise the annealing temperature to at least Tm + 3 when using another calculator's Tm. The specific rules the page quotes — Q5 at Tm_lower + 1 (stated in the calculator's own Q5 message), Phusion at 0.93 × Tm + 7.5, Taq buffers at Tm − 5 — are what the calculator displays, checked on five primer pairs against SeqBench's reimplementation (tests/polymerase-annealing.test.ts). DISAGREEMENT within NEB: its 'Guidelines for PCR Optimization with Thermophilic DNA Polymerases' page gives 0–2 °C below the lowest Tm for non-Q5/Phusion enzymes, where the calculator applies Tm − 5 to Taq, OneTaq and LongAmp; the page follows the calculator, which NEB's product protocols recommend.
- 4PCR cycling considerationsThermo Fisher ScientificThe Taq-type starting point in 'Setting the annealing temperature', the PROGRAM table's Annealing note and the FAQ: "A general rule of thumb is to begin with an annealing temperature 3–5°C lower than the lowest Tm of the primers" (fetched 2026-09-25). This is the source the page's 3–5 °C figure rests on; note that it names three Tm methods and does not tie the rule to any one, which is why the page tells the reader to take rule and Tm from the same supplier.
- 5Optimization of the annealing temperature for DNA amplification in vitroRychlik W, Spencer WJ, Rhoads RE · Nucleic Acids Research 18(21):6409-6412 · 1990The 'Setting the annealing temperature' section's two failure modes and its choice of WHICH primer's Tm to use. The abstract (fetched from PMC332522) states directly: "At both sub- and super-optimal Ta values, non-specific products may be formed, and the yield of products is reduced" — that is the page's too-high / too-low paragraph and the matching FAQ answer. It also states "The TaOPT is found to be a function of the melting temperatures of the less stable primer-template pair and of the product", which is the primary basis for the page saying to work off the LOWER of the two primer Tm values. Note what it does NOT support: Rychlik's actual result is a calculated optimum (a function of both primer and product Tm), not a fixed offset below primer Tm, so this paper does not back the specific "3-5 °C" figure — the Thermo Fisher entry above does, as a general rule of thumb, which the page now applies to Taq-type enzymes only. Abstract and record verified; the 4-page full text on PMC is page scans.
- 6'Touchdown' PCR to circumvent spurious priming during gene amplificationDon RH, Cox PT, Wainwright BJ, Baker K, Mattick JS · Nucleic Acids Research 19(14):4008 · 1991The primary source for touchdown PCR — both the 'Two-step and touchdown PCR' section and the touchdown FAQ answer. This is the paper that named and introduced the technique. Record verified at PMC328507 / PMID 1861999 and the DOI resolves. CAVEAT: it is a one-page note whose full text on PMC is a single page scan, so I could confirm the paper's identity and provenance but could not machine-read its stated step-down increment or cycle counts. The page's description of touchdown is generic ("starts a few degrees above the expected optimum and steps it down over the first several cycles") and does not quote a specific increment, so nothing numeric on the page depends on the unreadable text.