Choosing a Buffer for a Double Digest
9 min read · Updated September 12, 2026
Cutting a vector with two enzymes in one tube saves a cleanup step and a couple of hours, and most of the time it simply works. When it does not, the failure is usually quiet: a partial digest that looks like a successful one on a gel until the ligation comes back with a background of uncut vector.
This guide covers how to pick the buffer, what to do when no buffer is good for both, and the two things that will sink a double digest even when the buffer choice is perfect — incubation temperature and star activity.
Why one buffer is a compromise at all
Restriction enzymes differ mainly in the salt concentration they prefer. An enzyme that wants 100 mM NaCl and one that wants none will not both be at their best in the same tube, and the units on the label are defined at the enzyme's own optimum.
NEB's four-buffer system — rNEBuffer 1.1, 2.1, 3.1 and rCutSmart — spans that range, and every enzyme is characterised as a percentage of its maximum activity in each of the four. Picking a buffer for two enzymes therefore means finding the buffer that maximises the worse of the two activities, which is a small optimisation you can do by reading a table, and which the Double Digest Buffer tool does for you from two enzyme names.
The reason to maximise the minimum rather than the sum is that a double digest is only as good as its weaker enzyme. A pair running at 100 and 25 percent leaves you with vector that is fully linearised and only partly cut a second time — which is exactly the singly-cut, re-ligatable backbone that produces a plate of empty-vector colonies.
How often it just works
Taking the 49 enzymes SeqBench holds NEB reaction data for gives 1,176 possible pairs. Running every one of them through the buffer optimisation:
- 762 pairs (65%) have a buffer where both enzymes are at 100% — no compromise at all.
- 144 pairs (12%) top out at 75% for the weaker enzyme.
- 158 pairs (13%) top out at 50%.
- 112 pairs (10%) cannot get the weaker enzyme above 50% in any single buffer, and of those, 11 pairs have no buffer in which both enzymes work at all.
Reading those numbers
So the common case is genuinely easy, and the folklore that double digests are fiddly comes from the third of pairs where they are. EcoRI and BamHI, the pair in every cloning course, are both at 100% in rNEBuffer 2.1 and both incubate at 37 degrees — nothing to think about.
A 75% pair is not a problem either. Activity is a rate, and a digest is normally run well past completion, so 75% of maximum for an hour is still a complete digest. SalI with XbaI is an example: 75% in rNEBuffer 3.1, and in practice you would extend the incubation slightly and not think about it again.
At 50% the honest advice is to add more units rather than more time, because the things that go wrong with a long digest — star activity, exonuclease contamination, evaporation — get worse with time and not with units. At 25% or below, stop trying to do it in one tube.
When to go sequential
NotI and SmaI is the kind of pair that has to be split: the best single buffer leaves the weaker enzyme at 25%. The recipe for a sequential digest is the same every time, and the order is not arbitrary.
Cut with the enzyme whose optimal buffer has the lower salt first, in its own best buffer. Then heat-inactivate it, or clean up on a spin column if it does not heat-inactivate. Then adjust the salt upwards and add the second enzyme. Going in this direction means the second step is a salt addition rather than a dilution, so you are not fighting the reaction volume.
The heat-inactivation step is where people get caught, because not every enzyme can be inactivated by heat — BamHI is a common one that cannot, and neither can TaqI. For those, a column cleanup between steps is the only safe option; leaving an active first enzyme in the tube while you incubate the second for another hour is a good way to produce star activity that you then have to diagnose.
Temperature is a separate question, and it is easy to miss
The buffer table is not the whole compatibility question. Of the 1,176 pairs, 229 — nearly one in five — involve two enzymes with different incubation temperatures, and that is independent of whether the buffer works.
EcoRI with TaqI is the clean illustration. Both are at 100% in rNEBuffer 2.1, so a buffer table says the pair is perfect. EcoRI incubates at 37 degrees and TaqI at 65. There is no temperature at which both are doing what their units were measured at.
Most enzymes sit at 37, and the exceptions are a small enough list to memorise for the common set: SwaI at 25, ApoI at 50, BstYI at 60, and BstBI and TaqI at 65. When one of those is in your pair, the digest is sequential by temperature even if the buffer is shared — cut with the lower-temperature enzyme first, then raise the temperature and add the thermophilic one.
Star activity
Star activity is an enzyme cutting sequences that resemble but do not match its recognition site, and it turns a clean double digest into a smear of unexpected fragments. It is provoked by exactly the conditions a difficult double digest pushes you towards: long incubations, excess units, glycerol above about 5 percent of the reaction volume, high pH, and the wrong salt.
Eighteen of the 49 enzymes in the set are flagged for star activity, and eight of those are high-propensity: EcoRI, BamHI, HindIII, SalI, PstI, EcoRV, PvuII and ApoI. That list is close to a list of the most commonly used enzymes in cloning, which is not a coincidence — they are well characterised because they are heavily used.
The glycerol point is the one most often missed, because it follows from a unit calculation rather than being a separate decision. Enzymes are supplied in 50% glycerol, so adding more than a tenth of the reaction volume as enzyme puts you over 5% glycerol. If you find yourself pipetting 3 microlitres of each enzyme into a 20 microlitre reaction to compensate for a 50% buffer, you have solved the activity problem by creating a star-activity one. Scale the reaction volume up instead.
Methylation, which is not a buffer problem but looks like one
Plasmid DNA from a standard lab E. coli strain carries Dam and Dcm methylation, and some enzymes will not cut a methylated site. The symptom is an enzyme that appears to be dead, in a reaction where the other enzyme worked perfectly — which reads exactly like a buffer incompatibility and is not.
In the common set, XbaI, ClaI, BspEI, TaqI and MboI are blocked by Dam methylation, and StuI, ApaI, FseI and ScrFI by Dcm. Several others are blocked by CpG methylation, which matters for DNA from mammalian sources rather than from E. coli.
Dam and Dcm blocking is also context-dependent rather than absolute: XbaI is blocked only when its site happens to be overlapped by a Dam site, which depends on the neighbouring bases. So the same enzyme can cut one plasmid and not another, which makes it look intermittent. If an enzyme fails in a spot where it has worked before, check the flanking sequence before you blame the tube — and if it is blocked, the fix is to prepare the plasmid from a dam-/dcm- strain, not to change the buffer.
The order to check things in
- Do both enzymes have a site in your construct, and are the sites far enough apart to give fragments you can resolve? Cut positions first, buffer second.
- Is there a buffer where both are at 100%? Two thirds of the time, yes, and you are done.
- If not, is the weaker enzyme at 50% or better? Add more units, not more time.
- Below 50%, split into a sequential digest, low salt first, and check whether the first enzyme can be heat-inactivated.
- Do the two enzymes want the same temperature? This is a separate question from the buffer and sinks one pair in five.
- Is either enzyme a high-propensity star-activity enzyme, and will your enzyme volume push glycerol over 5%?
- Is either site blocked by Dam or Dcm methylation in the strain the plasmid came from?
Frequently asked questions
Which buffer should I use for a double digest?
The one that maximises the activity of the weaker enzyme, not the average of the two — a digest is only as complete as its worse enzyme. Across the 49 enzymes with NEB reaction data here, 762 of the 1,176 possible pairs (65%) have a buffer where both are at 100%. Enter the two enzyme names into the Double Digest Buffer tool to get the recommendation and the full four-buffer table.
What activity percentage is good enough for a double digest?
75% is fine — activity is a rate and a digest is normally run past completion, so a slightly slower enzyme still finishes. At 50%, add more units rather than extending the incubation, because long digests promote star activity while extra units do not. Below 50%, split it into a sequential digest.
How do I do a sequential digest?
Cut with the enzyme whose optimal buffer has the lower salt first, in its own best buffer. Heat-inactivate it, or clean up on a spin column if it cannot be heat-inactivated — BamHI and TaqI are common enzymes that cannot. Then raise the salt and add the second enzyme. Going low-salt first means the second step is an addition rather than a dilution.
Can I double-digest with two enzymes that need different temperatures?
Not in one step. Nearly one in five enzyme pairs has a temperature mismatch, and it is independent of the buffer — EcoRI and TaqI are both at 100% in rNEBuffer 2.1 but incubate at 37 and 65 degrees. Digest at the lower temperature first, then raise it and add the thermophilic enzyme. Most enzymes are 37; in the common set the exceptions are SwaI at 25, ApoI at 50, BstYI at 60, and BstBI and TaqI at 65.
What causes star activity in a double digest?
Long incubations, excess units, glycerol above about 5% of the reaction volume, high pH and the wrong salt — which is the list of things a difficult double digest tempts you into. Eighteen of the 49 enzymes here are flagged for it and eight are high-propensity: EcoRI, BamHI, HindIII, SalI, PstI, EcoRV, PvuII and ApoI. Enzymes ship in 50% glycerol, so keep total enzyme under a tenth of the reaction volume.
Why did one of my two enzymes not cut at all?
If the buffer was fine for both, suspect methylation before anything else. Plasmid from a standard E. coli strain carries Dam and Dcm methylation; XbaI, ClaI, BspEI, TaqI and MboI are blocked by Dam, and StuI, ApaI, FseI and ScrFI by Dcm. Blocking is context-dependent — XbaI is only blocked when a Dam site overlaps its own — so the same enzyme can cut one construct and not another. The fix is a dam-/dcm- prep, not a different buffer.
Is it ever better to just do two separate digests?
Yes, whenever the one-tube version needs compromises you would have to verify anyway. A sequential digest costs a cleanup and an hour; an incomplete double digest costs a transformation, a plate of empty-vector colonies and a week. The 10% of pairs that cannot get past 50% in any buffer are the clear cases, and a temperature mismatch is another.
Does it matter which enzyme I add first in a sequential digest?
Yes. Cut with the low-salt enzyme first so the second step raises the salt rather than requiring a dilution, and check that the first enzyme can be heat-inactivated before you plan to skip the cleanup. Leaving an active first enzyme in the tube through a second hour-long incubation is a common route to star activity.
Related references
Common restriction enzymes: recognition sites, cut positions, NEB buffer activity, star activity and an interactive double-digest buffer finder.
Nucleotide ambiguity codes and their complements.
Selection markers, mechanisms and working concentrations for cloning.
Related tools
Pick two restriction enzymes and get the single buffer that runs both, each enzyme's activity in every buffer, and a straight answer on when to digest sequentially instead.
Find recognition and cut sites for common restriction enzymes.
Predict restriction fragments and see the simulated agarose gel with a ladder.
Find the restriction digest that tells your intended construct apart from the empty vector, a flipped insert or a mis-assembly — with the bands you would actually be able to read.