Name the limit procedure first
This is the SD/slope approach in ICH Q2(R2), section 3.2.3.3, not the definition of every possible detection-limit procedure.[1]
DL denotes detection limit, often called LOD; QL denotes quantitation limit, often called LOQ. Factors 3.3 and 10 belong to this specified estimation approach. They do not automatically guarantee a particular error probability or constitute a universal three-sigma decision rule. The corrected ICH guideline also describes visual, signal-to-noise and direct accuracy/precision approaches. A differently defined limit may therefore differ numerically without that difference alone proving a calculation error.[1]
σ and S must use the same response scale and conditions relevant to the sample. Dimensionless normalized response with S in L/mg gives a limit in mg/L. For a decreasing calibration line, use positive sensitivity magnitude |b|; a negative concentration limit has no useful interpretation here. The analytical calibration line explains intercept and slope. Do not subtract the intercept from σ: σ describes variability rather than mean background.[1][2]
Which standard deviation is intended?
ICH lists possible sources for σ including variability of appropriate blank responses, residual standard deviation from a suitable low-range calibration, and standard deviation of intercepts from several regression lines. These are not interchangeable labels. Document data, repetitions, matrix and analysis. Residual variability determined far above the limit may not represent the lower end, especially when variance changes. The procedure must justify which quantity describes relevant response variability.[1]
The standard deviation of individual responses is not automatically the standard error of their mean, σ/√n. Replacing σ by a smaller standard error merely because more blank replicates were measured changes the limit’s statistical basis. Response SD is also different from variability already converted into concentration units; dividing that concentration variability by S again would count sensitivity twice. The calculator expects a positive, justified response SD and does not estimate it from a single number.[1][2]
An observed σ=0 in a short, rounded series does not establish a perfect zero analytical limit. It may reflect limited data, resolution or identical rounded outputs. For a usable estimate, the calculator requires σ>0 and S>0. Likewise, QL/DL=10/3.3 holds only when both values use the same σ/S approach; it is not a universal law for separately determined experimental limits.[1]
Two complete limit estimates
The existing example with explicit units
- Given
- An appropriate dimensionless response SD is σ=0.006. Matching sensitivity is S=0.20 L/mg. Both values are supplied for the same lower method interval.
- Find
- Calculate DL and QL in the declared ICH approach.
State both factors of the named SD/slope approach before substitution.
Calculate the shared concentration scale first.
Apply the two distinct factors.
Result: Estimated quantitation limit exceeds estimated detection limit. Both refer to this procedure and response variability.
Check: QL/DL=0.300/0.099=100/33≈3.0303. With S=200 m3/kg, DL=0.000099 kg/m3, the same concentration.
Negative calibration slope and sample dilution
- Given
- An assessed normalized response has b=−0.40 L/mg and appropriate SD σ=0.004. The measured solution is prepared by lossless fivefold dilution of an original sample determined on the same basis.
- Find
- Separate positive measured-solution limits from the original concentration basis.
First define positive slope magnitude and the two concentration bases.
Use slope magnitude.
Transfer the concentration basis using the declared factor.
Result: Decreasing response does not create negative limits. Original-sample concentration includes dilution factor when method performance matches that preparation.
Check: Dividing by five returns measured-solution limits 0.033 and 0.100 mg/L. If preparation changes matrix, recovery or variability, reassess the method basis.
Recall the method later with the formula’s flashcards.
10 flashcards · English materials
Already have an account? Sign in for flashcardsAvailable from Essential. A ready-made formula set, no AI credits required. Article exercises are not all included.
Estimate, assess and report separately
ICH recommends presenting the limit and supporting approach and data. Calculated or extrapolated DL estimates can be assessed using suitable independent samples near the limit. An estimated QL should subsequently be checked with suitable samples near that limit; the guideline describes justified exceptions, such as an estimate far below the reporting limit. Alternatively QL can be established directly through accuracy and precision at the lower end. Neither division automatically performs these experimental assessments.[1]
A back-calculated concentration below estimated DL does not prove absence of analyte. A concentration between DL and QL may provide a detection indication without supporting adequately performing quantitation. Even above QL, matrix, range and procedure must match. Reporting depends on the specified procedure and intended use; this article establishes no universal reporting, legal or clinical decision limit.[1]
Five different exercises
Calculate · Foundation
Given σ=0.005 and S=0.25 L/mg for the same normalized response and matrix, calculate both ICH estimates.
Show hint
The shared ratio σ/S has concentration units.
Learning goal: Calculate DL and QL with matching factors.
Show solution and calculation
σ/S=0.020 mg/L.
DL=3.3·0.020=0.066 mg/L and QL=10·0.020=0.200 mg/L.
Check: QL/DL=100/33. These are estimates from the declared approach.
Rearrange · Application
In the same ICH approach, QL=0.080 mg/L at S=0.50 L/mg. Which σ was used?
Show hint
Rearrange QL=10σ/S for σ.
Learning goal: Infer matching response SD from QL.
Show solution and calculation
σ=QL·S/10=0.080·0.50/10=0.0040, dimensionless.
Check: 10·0.0040/0.50=0.080 mg/L. The same approach gives DL=3.3·0.0040/0.50=0.0264 mg/L; inversion does not establish that σ was experimentally estimated appropriately.
Error analysis · Application
25 suitable blank replicates have response SD s=0.010. Sensitivity is S=0.50 L/mg in the ICH SD/slope approach. Someone substitutes s/√25=0.002 merely because there are more replicates. Compare DL values.
Show hint
Variability of individual responses differs from uncertainty of their mean.
Learning goal: Do not replace standard deviation by standard error.
Show solution and calculation
Using declared response SD gives DL=3.3·0.010/0.50=0.066 mg/L.
Unjustified standard-error substitution gives 0.0132 mg/L, five times smaller.
More replicates can improve estimation of s without automatically shrinking individual-response variability fivefold.
A different mean-response procedure would need its own explicit statistical justification.
Interpret · Application
A declared procedure has estimated DL=0.10 mg/L and QL=0.30 mg/L. A back-calculated value is 0.18 mg/L. Does this establish adequately accurate quantitation?
Show hint
Place the value between the limits and retain their estimated status.
Learning goal: Distinguish detection from quantitation.
Show solution and calculation
0.10<0.18<0.30 mg/L.
The value lies between estimated limits and does not automatically establish adequately performing quantitation.
Reporting a detection indication requires procedure rules and suitable evidence.
Inferring zero analyte would also be wrong.
These rounded supplied limits need not have exactly the ratio 10/3.3.
Transfer · Challenge
Method A has σA=0.006 and SA=0.20 L/mg; method B has σB=0.009 and SB=0.60 L/mg. Both use the same declared approach and appropriate matrix data. Compare estimated DL without claiming overall method superiority.
Show hint
Compare σ/S rather than σ or S alone.
Learning goal: Balance sensitivity gain against greater response variability.
Show solution and calculation
A: σ/S=0.030 mg/L and DL=0.099 mg/L.
B: σ/S=0.015 mg/L and DL=0.0495 mg/L.
Despite greater response SD, tripled sensitivity halves estimated DL here.
Check: (σB/σA)/(SB/SA)=1.5/3=0.5. This does not establish accuracy, selectivity, robustness or cost superiority.
Check your understanding with the formula’s reference test.
8 test questions · English materials
Already have an account? Sign in for the testAvailable from Essential. A ready-made formula set, no AI credits required. Article exercises are not all included.
Sources for this explanation
Quanta · Updated November 1, 2026
- ICH Q2(R2), corrected 4 April 2025, sections 3.2.3.2–3.2.3.5
- Harvey, Chemometrics Using R, chapter 8 Modeling Data, response and calibration
Editorial responsibility: Quanta Study editorial team
Review scope
This explanation, 2 worked examples and 5 exercises are available in German and English. Structure, source references and mathematical notation undergo automated checks. Calculations, model assumptions and both language versions receive a separate AI review.
An independent, named human subject review has not yet been documented. Our exercise numbers are worked examples, rather than measurements published by the cited sources.
Editorial responsibility, sources and methodsChange history
- Added current ICH sections, choice of σ, slope magnitude, dilution basis and validation limits with two examples and five exercises.
Cite this reference page
Quanta. Limit of Detection and Limit of Quantification. Updated November 1, 2026. https://quanta-study.de/en/formula-sheet/nachweis-und-bestimmungsgrenze
When citing a specific scientific statement, also cite the specialist source linked to that statement.