Technique hub
ELISA protocols — versioned, with troubleshooting and fixes.
An ELISA (enzyme-linked immunosorbent assay) quantifies a target by linking it to an enzyme that turns a substrate into a measurable signal on a plate reader. Most failures cluster in a few places: high background from under-blocking or weak washing, no signal from a dead conjugate, and standard curves that saturate or scatter. The ELISA protocols below keep those fixes visible — version by version.
Sandwich ELISA for cytokine quantification
Indirect ELISA for antibody titer
Direct ELISA with HRP-conjugated antibody
Competitive ELISA for haptens
Four-parameter (4PL) standard curve fitting
High background & no-signal rescue
Step by step
A sandwich ELISA, step by step — and where each one fails.
The classic eight-step sandwich ELISA. Vendor pages list the steps; what they leave out is the failure that lives in each one. We keep both together.
Coat the plate
Adsorb the capture antibody (or antigen) onto a high-binding plate in carbonate–bicarbonate coating buffer, overnight at 4 °C.
Block
Wash, then saturate the remaining surface with a protein blocker (BSA, casein, or a commercial blocker) for 1–2 h.
Add samples & standards
Load diluted samples and a serial-dilution standard series in replicate; reserve wells for blanks.
Add detection antibody
Apply the detection (and, for indirect formats, secondary) antibody at its titred dilution.
Add enzyme conjugate
Bind the enzyme conjugate (e.g. streptavidin–HRP) that turns binding into a measurable signal.
Wash thoroughly
Wash 3–5× with PBS/TBS + 0.05 % Tween-20, with full aspiration between cycles.
Develop the substrate
Add the chromogenic substrate (commonly TMB) and let colour develop in the dark, watching the standards.
Stop & read
Stop the reaction (e.g. 2 N H₂SO₄) at a fixed time and read optical density on a plate reader.
Troubleshooting
ELISA troubleshooting, grouped by what went wrong.
Five failure families cover almost every ELISA that doesn't work the first time. Match the symptom, find the likely cause, apply the fix.
High background
| Symptom | Likely cause | Fix |
|---|---|---|
| Whole plate reads high, blanks included | Insufficient blocking; blocker incompatible with the analyte | Extend block to 2 h; switch BSA ↔ casein; match blocker to the conjugate |
| Background creeps up well-to-well | Inadequate washing / Tween-20 left out | Add 0.05 % Tween-20 to all washes; raise wash count to 4–5× with full aspiration |
| Background only in detection-antibody wells | Detection or secondary antibody too concentrated / cross-reactive | Titre the antibody down; pre-adsorb the secondary against the sample matrix |
No signal / no colour development
| Symptom | Likely cause | Fix |
|---|---|---|
| No colour anywhere, standards included | Dead substrate or inactivated enzyme conjugate | Test substrate with fresh conjugate; check conjugate storage and expiry |
| Signal in standards but not samples | Analyte absent, degraded, or masked by matrix | Confirm sample handling; spike a known amount to check recovery |
| Colour develops then fades fast | Stop solution wrong or added too late | Use the substrate-matched stop reagent; fix a consistent develop time |
Weak / low signal
| Symptom | Likely cause | Fix |
|---|---|---|
| Top standards barely above blank | Over-diluted detection antibody or conjugate | Re-titre antibody and conjugate; verify coating concentration |
| Whole assay shifted low | Coating failed or develop time too short | Check coating buffer pH; lengthen develop and watch the standards |
Standard-curve problems
| Symptom | Likely cause | Fix |
|---|---|---|
| Curve flattens at the top (saturation) | Over-development or standards above the assay range | Shorten develop time; narrow the standard range; fit with 4PL, not a line |
| High replicate CV (>15 %) | Pipetting, edge effects, or uneven washing | Use reverse pipetting; avoid edge wells or buffer-fill them; standardise washing |
| Low-end points hug the blank | Blank not subtracted; assay near its detection limit | Subtract the blank; report values below LLOQ as <LLOQ, don't extrapolate |
Edge effects / well-to-well variability
| Symptom | Likely cause | Fix |
|---|---|---|
| Outer wells read differently from the centre | Thermal/evaporation gradient during incubation | Pre-warm reagents; cover the plate; avoid the outermost ring for samples |
| Random scatter with no pattern | Bubbles, incomplete aspiration, or reader artefacts | Tap out bubbles; verify full aspiration; blank-read the empty plate |
The moat
Every fix stays attached to the protocol.
The reason a blocker was switched, the develop time that stopped the curve saturating, the wash tweak that halved your CV — on jnar that knowledge lives in the protocol, version by version. Not in someone's notebook, not in a folder of elisa_v3_FINAL.docx copies.
- ✓What changed and why, kept side by side
- ✓Negative results recorded as first-class data
- ✓Every change attributed via ORCID
In depth
ELISA standard curve problems, fixed properly.
An ELISA standard curve is sigmoidal, not linear. Fitting a straight line through the middle and extrapolating to the ends is the most common reason quantification goes wrong — it distorts both the low and high reaches of the curve. Fit a four-parameter logistic (4PL) model instead; use 5PL when the curve is visibly asymmetric.
Keep your standards inside the working range so the top points don't saturate — a flat top means the assay has run out of dynamic range, and shortening the develop time usually pulls it back. At the bottom, subtract the blank before fitting and treat anything under the lowest reliable standard as <LLOQ rather than extrapolating into noise.
Watch the hook (prozone) effect: very high analyte concentrations can paradoxically lower the signal, folding the curve back on itself. If a sample reads improbably low, dilute it and re-run before trusting the number. And track replicate CV — above ~15 % the problem is usually pipetting, edge effects, or uneven washing, not the curve fit.
Looking for an ELISA protocol PDF?
A PDF goes stale the moment someone changes a step. Import your methods PDF into jnar instead and get a structured, versioned ELISA protocol with a living fix history — the changes and their reasons stay together. Your first AI conversion is free.
FAQ
ELISA troubleshooting, answered.
Why is my ELISA background so high?
High background is almost always one of three things: under-blocking, inadequate washing, or a detection antibody that's too concentrated or cross-reactive. Extend the block, add 0.05 % Tween-20 to every wash with full aspiration, and titre the detection (and secondary) antibody down. If the background tracks the conjugate even in blank wells, suspect contamination or a blocker that doesn't match the conjugate.
Why is there no signal or no colour development in my ELISA?
If even the top standards stay colourless, the substrate or the enzyme conjugate is dead — test the substrate against a fresh conjugate and check storage and expiry. If the standards develop but the samples don't, the analyte is likely absent, degraded, or masked by the sample matrix; spike a known amount to check recovery. Colour that appears and then fades usually means the wrong stop solution or a develop time that drifted.
How do I fix ELISA standard curve problems?
Fit a four-parameter logistic (4PL) curve rather than a straight line — ELISA response is sigmoidal, so a linear fit distorts both ends. Keep your standards inside the assay's working range so the top points don't saturate, subtract the blank before fitting, and report anything below the lowest reliable standard as <LLOQ instead of extrapolating. High replicate CV usually traces back to pipetting, edge effects, or uneven washing.
What are the steps of a sandwich ELISA?
Coat the plate with capture antibody, block the remaining surface, add samples and a standard series, add the detection antibody, add the enzyme conjugate, wash thoroughly, develop the chromogenic substrate, then stop and read optical density on a plate reader. Each step has a characteristic failure mode — under-blocking causes background, pipetting errors wreck the standard curve, and inconsistent stop timing hurts reproducibility.
What's the difference between direct, indirect, sandwich, and competitive ELISA?
Direct ELISA uses one enzyme-labelled antibody against an immobilised antigen — fast but less sensitive. Indirect adds a labelled secondary antibody for amplification (common in serology). Sandwich captures the antigen between two antibodies and is the workhorse for quantifying proteins like cytokines. Competitive ELISA measures signal that drops as analyte increases, which suits small molecules and haptens that can't be sandwiched.
Do you have an ELISA protocol PDF I can download?
Rather than a static PDF that goes stale the moment someone tweaks a step, jnar lets you import your own ELISA protocol — drop in a methods PDF and turn it into a structured, versioned protocol with a visible fix history. You keep the changes and the reasons together instead of in a file nobody updates. Your first AI conversion is free at /import.
Have an ELISA method of your own?
Import it into jnar and start recording the fixes. Your first AI conversion is free.