Products & safety

Laboratory detection limits: non-detects, quantification, uncertainty and what a product test can tell you

6 sources · Latest source check 2026-10-13 UTC

On this page

A laboratory detection limit describes a measurement method’s ability to distinguish a small signal from background. “Not detected” is therefore a result with a method and threshold attached, not proof of absolute absence. A report should identify the chemical, sample, units and relevant limit so that readers can interpret it.[1]

Detection, quantification and reporting limits differ

Detection distinguishes a signal from background; quantification assigns an amount with defined performance. A reporting limit can reflect a program’s reporting convention. EPA’s water-data guide distinguishes non-detected results from estimated values between detection and quantitation levels, while noting that terminology varies. “Below reporting limit” should retain its qualifier and threshold rather than being read as zero.[1]

A method limit includes more than instrument sensitivity

EPA’s named MDL procedure defines a confidence-based distinction from method-blank results and requires the sample-processing steps to be included. A method blank is processed without the target sample so background contamination can be characterized. Spiked samples contain added known amounts. The procedure compares blank- and spike-based estimates. A very sensitive instrument alone does not guarantee an equally sensitive result after extraction, dilution, digestion or other preparation; the full method and sample context must match.[2]

A product name is not enough to define the tested sample

Eurachem’s validation guidance emphasizes fitness for purpose, sampling, handling, performance characteristics and documentation. That framework makes the intended question explicit: one container, one batch, a composite of several units, and repeated batches support different inferences. Matrix means the material containing the target chemical; food, water and a coating are different matrices. This is an editorial application of the guidance, not a claim that the Atlas has validated a method. A result for one sample cannot automatically certify every batch.[3]

How two non-detect reports can mean different things

Illustratively, “not detected below 1 unit” and “not detected below 10 units” use different thresholds; the latter does not establish absence at the former. Preserve units and qualifiers. EPA’s guidance similarly treats censored results differently from ordinary measured concentrations.[1]

A pass label needs a requirement and a decision rule

ILAC describes decision rules that account for uncertainty when stating conformity with a requirement. Simple acceptance and guard-banded acceptance can produce different judgments near a limit. Guard bands adjust the acceptance boundary to address decision risk; they are not new toxicity thresholds. A legal or specified rule may control which approach applies. “Passed” should identify the applicable requirement, results and rule. It does not automatically mean no contaminant, no risk at all or suitability for every use.[5]

Laboratory competence supports reliability but has a scope

ISO/IEC 17025 concerns competence, impartiality and consistent laboratory operation. ISO says accreditation bodies use it when assessing laboratories. This supports confidence in an appropriate testing service, while the actual report still needs its method, sample identity, result and qualifiers. A logo is not a substitute for checking whether the requested test and matrix are included in the laboratory’s relevant accredited scope. This page explains interpretation and does not independently audit or endorse a laboratory.[6]

Evidence & uncertainty

Three questions should remain separate: was a substance detected, was its concentration quantified reliably, and does the result meet a specified requirement? A pass/fail decision may depend on measurement uncertainty and a documented decision rule, not merely a number’s position on a chart.[5]

Selected-source overview · Independent review pending

Published questions

Does ND mean exactly zero?

Selected-source overview · Independent review pending. Read the evidence context.

No. It means the report did not establish detection at its stated threshold and under its method. Preserve the qualifier and units.[1]

Evidence behind this answer

WQX User Guidance: Detection Limits
agency analytical reporting guidance · 2025
Original official PDF selected reporting definitions and examples inspected; water-data guidance, broader examples are labeled editorial interpretation · Pages 2–8: non-detect conditions, detection/quantitation/reporting distinctions, units and qualifiers