ELISA Standard Curve

Fits a four-parameter logistic (4PL) through your standards and reads each sample's concentration back off it.Paste your standards and samples below.

One standard per line: known concentration first, reading second, separated by a tab, comma or spaces. Keep the zero standard in — it is the blank, and the fit uses it as the lower plateau. A header row is skipped.

Name, reading, and optionally the fold dilution. ⚠️ Separate the columns with a tab, comma or two spaces — a single space would make “Sample 1” into a number. A line holding only a reading works too.

How to use it

  1. Paste the standard series into the first box — known concentration in the first column, reading in the second. Keep the zero standard: it is the blank, and the fit uses it as the lower plateau.
  2. Paste the samples into the second box — name, reading, and the fold dilution if the sample was diluted before the well. Separate the columns with a tab, comma or two spaces.
  3. Read the concentrations off the table. A dilution is already multiplied in, with the concentration in the well shown underneath it.
  4. Check the curve below the table. The standards should sit on the line at both ends; where they do not, that plateau is guesswork and every concentration near it inherits the guess.
  5. Rows saying the reading fell off the curve got no number on purpose. Dilute those samples and read them again.

💡 The units are yours. Whatever the standards are given in, the samples come out in.

Formula & notes

An immunoassay does not answer in a straight line. It is flat at the bottom where there is nothing to detect, steep through the middle, and flat again at the top where the antibody is saturated — an S, and fitting a line to it is wrong everywhere except in the middle.

The four-parameter logistic is that S, written down. A is the reading at zero concentration, D the reading where the assay saturates, C the concentration half way between them, and B how steeply the curve climbs through C.

Standard curve
y = D + (A − D) / (1 + (x / C)^B)

Reading a sample is the same equation rearranged. It is exact — there is no search and no approximation.

Concentration from a reading
x = C × ( (A − y) / (y − D) )^(1/B)

The bracket has to be positive, and that is the whole safety story: it is positive exactly when the reading lies strictly between A and D. A reading outside them has no concentration — not a large one, not a small one, none — and the arithmetic refuses rather than rounding to the nearest plateau.

The reported concentration is what was in the original sample, so a dilution multiplies back in.

With a dilution
concentration = x × dilution factor

⚠️ The 4PL is symmetric about its inflection. Assays whose upper and lower halves have genuinely different shapes are fitted with a five-parameter logistic instead, which this tool does not offer — a fifth parameter that nobody knows when to turn on does more harm than a limitation stated plainly.

Practical notes

  • Run the standards on the same plate as the samples. A curve carried over from yesterday carries yesterday's coating, substrate and incubation with it.
  • Keep the blank in as the zero standard rather than subtracting it first. The lower plateau is the blank, estimated from the whole series instead of from one well; subtracting it and then fitting one counts it twice.
  • Replicate wells go in as separate rows. A fit told about both wells is better informed than a fit told about their mean, and two rows that disagree are worth seeing.
  • A reading beyond a plateau has no concentration at all — not a large one. Dilute the sample and read it again; the dilution column puts the answer back on the original scale.
  • Extrapolated is not the same as impossible. A concentration below the lowest standard or above the highest does exist on the curve, but no calibrator was ever measured there, so the number is the curve's opinion.
  • A high R² says the curve passes near the points. It says nothing about whether the plateaus were reached, and the plateaus are what the back-calculation divides by.

FAQs

Often used together