Western Blot Blocking Buffer: Choosing BSA or Milk by Target

Every western blot protocol tells you to block for an hour in 5% milk or 5% BSA, and most treat the two as interchangeable. They are not. The blocking agent is a competing protein sitting in the same tube as your antibody and, for several hours, in the same tube as your detection conjugate — so it fails in specific, predictable, chemistry-dependent ways. Non-fat dry milk fails because of two molecules it contains. BSA fails because of what its grade does or does not exclude. This page makes the choice by target and detection chemistry rather than by habit, and names the cross-reaction behind each rule.

What the blocking step actually has to do

After transfer, the membrane’s unoccupied surface still binds protein avidly and largely non-specifically — hydrophobically on PVDF, by a mix of hydrophobic and electrostatic interaction on nitrocellulose. That is the same property that captured your transferred proteins, and it does not distinguish between your antigen and the immunoglobulin you are about to add. Blocking floods the residual sites with an irrelevant protein so that later antibody binding is driven by affinity rather than by the membrane.

A blocker therefore has five jobs, and only two of them separate milk from BSA:

1. Bind the free membrane surface efficiently and stay bound through the washes.
2. Not bind your primary or secondary antibody.
3. Contain nothing that participates in your detection chemistry.
4. Not mask the epitope you are trying to detect.
5. Behave the same way from lot to lot.

Requirements 2 and 3 are where the real decision lives. On requirement 1, milk is usually the stronger blocker; on requirement 5, BSA is usually the more consistent one. Everything else is a tie until you name the antibody and the conjugate.

Non-fat dry milk: cheap, strong, and chemically busy

Non-fat dry milk blocks well precisely because it is a complex undefined mixture — caseins, whey proteins, immunoglobulins, glycoproteins, residual lipid, lactose and vitamins. Every one of those components can also do something you did not ask for.

Casein competes with phospho-specific antibodies

Casein is the dominant protein fraction of bovine milk — roughly 80% of total milk protein — and it is a phosphoprotein. The α_s1-, α_s2-, β- and κ-casein chains carry phosphate covalently esterified to serine residues (phosphoserine), clustered in the calcium-binding regions that give casein micelles their structure. A phospho-specific antibody is raised to recognise a phosphorylated residue in a defined sequence context, and the greater the share of its binding energy that comes from the phosphate group rather than the flanking sequence, the more readily it engages phosphoserine displayed on membrane-bound casein.

That produces two effects on the same blot, and the second is the one people misdiagnose:

• Background rises, because primary antibody binds casein spread evenly across the entire membrane.
• Specific signal falls, because casein in the antibody diluent titrates the antibody out of solution before it ever reaches the target band. This reads exactly like a weak antibody, a poor transfer or a low-abundance target — see the no-bands decision tree for how to rule those out first.

Not every phospho-antibody is affected; it depends on how phosphate-dependent that particular clone’s binding is. You cannot predict which clones will be, which is why the general rule is to avoid milk on phospho-blots rather than to test it each time.

Endogenous biotin breaks streptavidin–biotin detection

Milk contains biotin (vitamin B7) as a natural constituent. Streptavidin binds biotin with an affinity in the femtomolar range — among the strongest non-covalent interactions known in biology — and that strength is exactly what makes the contamination fatal rather than merely inconvenient. On a milk-blocked membrane, streptavidin–HRP binds biotin left on the membrane surface, while free biotin carried over into the antibody diluent occupies the conjugate’s binding sites before it can reach your biotinylated probe. You get high diffuse background and suppressed specific signal simultaneously, which is a confusing combination if you are looking for a single cause.

The same applies to every biotin-mediated amplification scheme: a biotinylated secondary followed by a streptavidin conjugate, or an avidin–biotin complex format. This is not a subtle interference — it is direct competition between your analyte and a blocker constituent for the identical binding site.

The other four milk problems

• Bovine immunoglobulin. Non-fat dry milk contains bovine IgG. An anti-bovine secondary will bind it, and so will an anti-goat or anti-sheep secondary that has not been cross-adsorbed against bovine serum proteins — bovids share substantial immunoglobulin homology. Check the cross-adsorption statement on the secondary’s datasheet before assuming this does not apply to you.
• Glycoproteins. Milk is glycoprotein-rich, so a milk block is a poor pairing with lectin blotting or any glycoprotein-directed detection.
• Endogenous phosphatase. Milk carries phosphatase activity, which is why milk is a poor block for alkaline-phosphatase-conjugated secondaries and one reason HRP conjugates dominate on milk-blocked blots.
• Lot and brand variability. Non-fat dry milk is a grocery commodity, not a reagent with a specification sheet. Brands and lots differ in casein content, residual fat and biotin. A result that reproduced last year against a different tub is not a controlled comparison, and “it used to work” is a real, common, and completely invisible source of drift.

One further point specific to fluorescent western systems: milk contributes autofluorescence, most consequentially in the shorter (~700 nm) near-infrared channel that near-IR imagers read. Manufacturers of near-IR detection systems sell fluorescence-compatible blockers specifically because of this; if you are imaging on a near-IR platform, take the blocker recommendation from your instrument’s own bulletin rather than from a chemiluminescence protocol.

BSA: defined, neutral, and only as clean as its grade

Bovine serum albumin is a single, well-characterised serum protein of about 66 kDa. It is not glycosylated and not phosphorylated, which is precisely why it is the standard substitute wherever casein or milk biotin is the problem. What it is not is a universally safer choice.

• Grade is not a detail. Generic “BSA, fraction V” is a crude ethanol-fractionation product, not a purified blocking reagent. Depending on grade it can carry residual bovine IgG (the same secondary cross-reaction problem milk has), protease activity, and fatty acids.
• BSA can contain biotin. This is the trap in the standard advice. BSA is a serum product, and serum contains biotin; “blocking grade” and explicitly biotin-free BSA are sold as distinct products because ordinary BSA is not reliably biotin-free. On a streptavidin-probed blot, “use BSA instead of milk” is only correct if the BSA is specified biotin-free on its certificate of analysis.
• It is not always the better blocker. On some membrane and antibody combinations BSA blocks less completely than milk and gives more background, not less. Switching to BSA and finding background got worse is a normal result, not a mistake.
• Cost. BSA costs several times what milk powder does per litre of buffer. For a lab running blots daily this is a real budget line, and it is the honest reason milk remains the default where it is chemically safe.
• Epitope masking still applies. BSA does not exempt you from over-blocking; an unusually long or concentrated block can mask the epitope with either agent.

Select by target and detection chemistry

The decision is made by what you are detecting and what conjugate reads it out, not by which blocker the last person in the lab used.

The block is only half of it — the antibody diluent matters more

This is the rule most protocols state too weakly, and the one most bench practice gets wrong. The blocking step is a 30–60 minute exposure. The primary antibody incubation is frequently overnight. Whatever protein sits in the diluent is in contact with your antibody an order of magnitude longer than the blocker ever was, under conditions that favour equilibrium binding.

So if you correctly switch to BSA for a phospho-blot but then dilute the primary in milk because that is what the protocol says, you have not fixed the problem — you have moved casein out of a one-hour exposure and into a sixteen-hour one. The equivalent error on a streptavidin blot puts free milk biotin directly into solution with the conjugate.

Match the blocker in the diluent to the blocker in the block, unless the antibody datasheet directs otherwise. A substantial number of commercial phospho-specific antibodies specify BSA in TBST for both steps for exactly this reason, and the datasheet is the more authoritative document.

The deliberate exception: block in milk, probe in BSA

There is one documented case for mismatching them on purpose. Cui and Ma, writing in BioTechniques in 2018, describe a sequential approach for streptavidin-probed western blots: block with milk, which blocks more completely and more cheaply, wash thoroughly to remove carried-over milk biotin, then run the streptavidin-conjugate incubation in BSA. The logic is to use milk’s blocking performance where it helps and remove its biotin before the step where it hurts.

Attribution note: this is a short Benchmarks-format report and its full text was not independently retrieved for this page, so take the wash stringency and concentrations from the paper itself rather than from this summary. Treat it as a technique worth knowing rather than as a substitute for biotin-free BSA if background persists.

Concentration and time: ranges, and what actually sets them

These are widely published protocol conventions, not universal recipes. Blocking conditions are antibody-, target- and system-specific, and any single figure quoted as “the” recipe is a starting point that someone else titrated against a different antibody.

• Milk: 3–5% (w/v) in TBST. 5% is the usual starting point; drop to 3% or lower if you suspect epitope masking or signal suppression.
• BSA: 1–5% (w/v) in TBST. 5% is common for phospho-blots; 1–3% is frequently sufficient and considerably cheaper.
• Detergent: 0.05–0.1% (v/v) Tween-20 in the block and washes. More detergent lowers background and also lowers specific signal — it is a trade, not a free improvement.
• Time and temperature: approximately 1 hour at room temperature with agitation, or overnight at 4 °C. Overnight blocking is not automatically better; over-blocking can mask the epitope and is a documented cause of unexpectedly weak bands.

What actually determines your numbers, in priority order:

1. The antibody datasheet. It was titrated against something real. Start there and deviate deliberately.
2. Target abundance. A low-abundance target tolerates less blocking stringency, because you cannot afford the signal you lose.
3. Membrane. PVDF’s higher protein-binding capacity generally needs a more complete block than nitrocellulose for the same antibody.
4. Substrate sensitivity. A femtogram-class ECL substrate makes background matter far more than a standard substrate does; the same blot can be acceptable on one and unusable on the other.
5. Buffer base. Use TBS-based buffers with AP conjugates, since phosphate inhibits alkaline phosphatase, and for phospho-work generally, where a phosphate-containing buffer is an avoidable complication.

Failure-mode map: which blocker choice causes what

If the symptom is not on this table, the cause is probably not the blocker: work through the no-bands decision tree, which splits transfer-stage from detection-stage failures with a single Ponceau check, and confirm the transfer chemistry itself against the transfer buffer decisions.

How to test a blocker change so the answer means something

Change one variable. Cut the membrane after transfer, or run duplicate lane sets and split the membrane, so both blockers see the same transfer, the same antibody dilution, the same wash regime and the same exposure. Two blots run on different days at different exposures answer nothing, and the literature is full of blocker comparisons that are really transfer comparisons.

Keep a loading control on both halves. Record lot numbers for the blocker as well as the antibody. And write the outcome somewhere reusable — antibody, clone, RRID, lot, blocker, concentration, diluent, membrane, result — because a lab’s blocking knowledge is otherwise held entirely in whoever ran the assay last, and is re-derived from scratch every time that person leaves.

Frequently asked questions

Is BSA or milk better for western blot?

Neither is better in general; the choice is determined by your detection chemistry. Use non-fat dry milk by default for HRP chemiluminescent detection of a non-phosphorylated target. Use BSA when the primary is phospho-specific, when detection runs through biotin–streptavidin, when the conjugate is alkaline phosphatase, when you are detecting glycoproteins, or when you are imaging in the near-infrared.

Why can’t you use milk with phospho-specific antibodies?

Because casein, which is about 80% of milk protein, is itself a phosphoprotein carrying phosphoserine residues. A phospho-specific antibody can bind those residues on membrane-bound casein, raising background, and can be bound by casein in the diluent, reducing the antibody available to detect your target.

Why does milk interfere with streptavidin detection?

Milk contains endogenous biotin. Streptavidin binds biotin with roughly femtomolar affinity, so streptavidin conjugate is captured by biotin on the membrane and in the diluent before it can find your biotinylated probe — producing high background and weak specific signal at the same time.

Does BSA contain biotin?

It can. BSA is a serum-derived product and serum contains biotin, which is why biotin-free and blocking-grade BSA are sold as separate products. For a streptavidin-probed blot, confirm “biotin-free” on the certificate of analysis rather than assuming that any BSA solves the problem.

Can I block in milk and dilute the antibody in BSA?

As a general habit, no — mismatching them usually means the longer, more consequential exposure is happening in the wrong buffer. The documented exception is the sequential approach for streptavidin-probed blots reported by Cui and Ma (2018), where the milk block is deliberately washed out before a BSA-based conjugate incubation.

What percentage of BSA should I use for blocking?

Commonly 1–5% (w/v) in TBST, with 5% a frequent starting point for phospho-blots. Take the datasheet’s figure first if the antibody specifies one; 1–3% is often sufficient and materially cheaper.

Is casein the same as milk for blocking purposes?

Purified casein blocks strongly without whole milk’s biotin, bovine IgG, glycoprotein and lipid load, so it is a reasonable option for demanding low-abundance detection. It is still a phosphoprotein, so it is never appropriate for phospho-specific antibodies.

Can I reuse blocking buffer?

Don’t. Blocking solutions are protein-rich, support microbial growth, and after use carry protein eluted from the membrane. Make them fresh, keep milk solutions cold, and discard rather than storing them at room temperature between blots.

Should blocking buffer be made in TBST or PBST?

Either works for standard HRP detection. Use TBST when the conjugate is alkaline phosphatase, since phosphate inhibits AP, and prefer TBST for phospho-work to keep phosphate out of the system entirely.

Related reading

• Western blot protocol basics — the full eight-step walkthrough this decision sits inside
• Western blot transfer buffer: Towbin, methanol and SDS decisions — the chemistry of the step immediately before blocking
• Western blot no bands: troubleshooting decision tree — symptom-first diagnosis when the blot has already failed
• SDS-PAGE gel selection and troubleshooting — the separation upstream of transfer
• ELISA protocol basics, immunohistochemistry and immunofluorescence — the same blocking logic in three other immunoassay formats
• Bradford protein assay — quantifying total protein so lanes are loaded comparably
• Buffer and solution preparation — percentage-solution and detergent mechanics
• Loading controls and total-protein normalization — how the blot is quantified once blocking and probing are settled, and the linearity check that has to pass first
• Lab operations — the wider cluster this guide belongs to

Sources and verification notes

• Cui Y, Ma L. Sequential use of milk and bovine serum albumin for streptavidin-probed western blot. BioTechniques 65(3):125–126, 2018. doi:10.2144/btn-2018-0006 (PMID 30227743). Citation verified against PubMed; full text not independently retrieved.
• Johnson DA, Gautsch JW, Sportsman JR, Elder JH. Improved technique utilizing nonfat dry milk for analysis of proteins and nucleic acids transferred to nitrocellulose. Gene Analysis Techniques 1(1):3–8, 1984 — the origin of non-fat dry milk as a blotting blocker. This journal is not indexed in PubMed, so the citation could not be verified against a primary index for this page.
• Vendor technical literature. The blocking bulletins published by Bio-Rad, Thermo Fisher, Cytiva, Merck and the near-infrared imaging vendors block automated retrieval or have moved since last indexing, and none was independently fetched for this page. The concentration and incubation ranges above are therefore given as widely published protocol conventions rather than as any single vendor’s specification — consult your own supplier’s current bulletin, and your antibody’s datasheet, for system-matched figures before adopting them.

Blocking conditions are antibody-, target- and detection-system-specific. Nothing on this page is a validated recipe for your assay; it is the decision framework and the named cross-reactions behind it.