HPLC in Pharmaceutical QC: The Complete Practical Guide

Last reviewed: November 2026 by Dr. Farrukh Mehmood, Ph.D., 16+ years in pharmaceutical QC/QA

High-performance liquid chromatography is the instrument a QC analyst spends most of the working day with. Assay, related substances, dissolution, content uniformity, cleaning validation and stability samples all end up on it. This guide puts the working knowledge in one place: how the system works, how to choose a column and prepare a mobile phase, what system suitability means under the current harmonised USP <621>, how a method is validated under ICH Q2(R2), how to do the calculations by hand, and how to keep the data defensible in an inspection.

I have written it for analysts in their first five years and for pharmacy graduates preparing for a QC role. It is practical on purpose. Where a number comes from a guideline, I name the guideline. Where a number is only common industry practice, I say so, because auditors know the difference.

How HPLC works

A pump pushes a liquid (the mobile phase) at high pressure through a steel tube packed with fine particles (the column). A small volume of sample solution is injected into that stream. Each compound in the sample spends part of its time dissolved in the mobile phase and part of its time held on the surface of the particles (the stationary phase). A compound that is held more strongly moves more slowly and leaves the column later. A detector at the column outlet records each compound as a peak. The time at which a peak appears is its retention time, and the area under the peak is proportional to the amount injected.

Two modes matter in QC. (For the basic parts of an HPLC system and how to choose one, see Learn HPLC: principle, parts and types.)

• Reversed-phase. The stationary phase is non-polar (for example C18 bonded to silica) and the mobile phase is polar (water or buffer mixed with acetonitrile or methanol). Non-polar compounds are retained longer. Increasing the organic solvent makes peaks elute earlier. This is the mode for most drug substances and products.
• Normal-phase. The stationary phase is polar (bare silica, amino, cyano) and the mobile phase is non-polar (hexane with a polar modifier). It is used for some isomers, fat-soluble vitamins and compounds that are unstable in water.

Elution is either isocratic (mobile phase composition constant throughout the run) or gradient (composition changes during the run, usually increasing organic content). Assay methods are often isocratic. Related substances methods are often gradient, because impurities cover a wide range of polarity.

System components and what goes wrong in each

Two volumes are worth understanding. Dwell volume is the volume between the point where solvents mix and the head of the column. It delays the gradient reaching the column and is the usual reason a gradient method behaves differently on two instruments. Extra-column volume is the tubing, injector and detector cell volume outside the column. Too much of it broadens peaks, which matters most with short, narrow columns and small particles.

Instrument qualification (IQ, OQ, PQ) and periodic calibration of flow rate, injector precision, oven temperature, wavelength accuracy and detector linearity sit under USP <1058> Analytical Instrument Qualification and your site SOP. A system that is out of calibration cannot be used for release testing, even if system suitability passes.

Column selection

A pharmacopoeial monograph tells you the packing by USP “L” code and gives the dimensions and particle size. For an in-house method, the validated method names the exact column. Your job is to use an equivalent column, not a similar-looking one.

What the dimensions mean

• Length. Longer columns give more plates and more resolution, with longer run time and higher pressure. 250 mm and 150 mm are the common lengths in older monographs; 50 to 100 mm with small particles is common in newer methods.
• Internal diameter. 4.6 mm is standard for HPLC. Narrower columns (2.1 or 3.0 mm) use less solvent and give more sensitivity for the same injected mass, but are less tolerant of extra-column volume.
• Particle size. 5 µm is conventional, 3 to 3.5 µm gives higher efficiency, and sub-2 µm particles need a UHPLC system because of the pressure.
• Carbon load, end-capping and pore size. Two L1 columns from different manufacturers can give different selectivity. Peak shape for basic drugs depends heavily on end-capping and silica purity. Pore size of about 100 Å suits small molecules; 300 Å is for peptides and proteins.

Column care rules I enforce in every lab

• One logbook per column: serial number, method or product, date, analyst, injections, pressure and plate count at first use.
• Dedicate columns to a method or product family where possible. Never share a column between ion-pair methods and other methods.
• Stay inside the manufacturer’s pH range. For ordinary silica-based phases this is roughly pH 2 to 8.
• Never leave buffer in a column. Flush with water–organic (no buffer) and then store in the solvent the manufacturer recommends, usually a high-organic mixture.
• Change mobile phases gradually. Going straight from buffer to 100% organic can precipitate salt inside the column.

Mobile phase preparation

More failed runs come from the mobile phase than from the instrument. Prepare it the same way every time, and record it so that someone else could repeat it exactly.

Step by step

• Use the right grades. HPLC-grade or gradient-grade solvents, freshly drawn purified water of HPLC quality, and analytical-grade buffer salts. Record lot numbers.
• Weigh, do not estimate. Calculate buffer salt mass from molarity and molar mass, allowing for hydrates. The molarity calculator does this for you.
• Adjust pH on the aqueous portion only, before adding organic solvent, unless the method states otherwise. A pH reading in a water–organic mixture is not the same quantity. Calibrate the pH meter with buffers that bracket the target.
• Measure volumes separately, then mix. Water and organic solvents contract when mixed, so 500 mL plus 500 mL is not 1,000 mL. “Mix 600 volumes of buffer with 400 volumes of acetonitrile” means measure each in its own cylinder.
• Filter aqueous buffers through a 0.45 µm membrane (0.2 or 0.22 µm for UHPLC and sub-2 µm columns). Choose a membrane that is compatible with the solvent.
• Degas by online degasser, sonication or vacuum filtration. Do not sonicate so long that volatile components are lost.
• Label with composition, pH, preparation date, expiry, analyst and reference to the preparation record.

A buffer controls pH properly only within about one pH unit of its pKa. Phosphate at pH 5 is a salt solution, not a buffer. For an ionisable drug, set the mobile phase pH at least 1.5 to 2 units away from the drug’s pKa where the method design allows. Close to the pKa, a small pH error produces a large retention shift.

Solvent choice affects detection. Acetonitrile has a UV cut-off near 190 nm and methanol near 205 nm, so low-wavelength methods (200 to 220 nm) normally use acetonitrile. Mixing water with methanol releases heat and mixing water with acetonitrile cools the mixture; in both cases let the mobile phase reach room temperature before use, or retention will drift during the first injections.

System suitability under USP <621>

System suitability testing (SST) confirms that the complete system (instrument, column, mobile phase, analyst and conditions) is performing adequately at the time of analysis. It is not a substitute for method validation or instrument qualification. It is the gate before sample data are accepted.

The current USP General Chapter <621> Chromatography is the text harmonised with Ph. Eur. 2.2.46 and JP 2.01 through the Pharmacopoeial Discussion Group. It became official on 1 December 2022, and its new peak symmetry and system sensitivity requirements, which USP had postponed, became official on 1 May 2025. The monograph always takes priority: if it gives a system suitability requirement, that is the requirement.

The parameters and how they are calculated

Older methods and some CDS reports use the tangent-width formulas, N = 16 × (t_R ÷ W)² and R_S = 2 × (t_R2 − t_R1) ÷ (W₁ + W₂). They give slightly different numbers. Use the calculation the monograph or validated method specifies and make sure the CDS is configured for it.

System repeatability limits

For the assay of an active substance or excipient where the target value is 100% for the pure substance, the harmonised chapter sets a maximum %RSD that depends on the number of injections and on B, the upper limit of the assay range in the monograph minus 100%.

So a substance with an assay range of 98.0 to 102.0% (B = 2.0) and six standard injections must give an RSD of not more than 0.85%. This table does not apply to related substances tests.

For finished products, the monograph or validated method states the limit. Not more than 2.0% for replicate standard injections is the figure most assay methods use. The 1994 FDA Reviewer Guidance “Validation of Chromatographic Methods” is still widely quoted for general targets: k greater than 2, resolution greater than 2, tailing factor not more than 2, plate count greater than 2,000 and RSD not more than 1% for five or more injections. Those are recommendations, not pharmacopoeial requirements. Check your own RSD by hand at least once with the RSD calculator.

Adjustments USP <621> allows without revalidation

A pharmacopoeial method may be adjusted within the limits below to meet system suitability. Anything beyond them is a modification that needs validation. After any adjustment, system suitability must pass, and for a pharmacopoeial method you should document a verification.

The stationary phase chemistry may not be changed: a C18 cannot be replaced by a C8. The chapter also permits a move from totally porous to superficially porous particles. In that case the L/dp rule is replaced by an efficiency check: the plate number (isocratic) or (t_R ÷ W_h)² (gradient) must stay within −25% to +50% of the original.

How I expect a sequence to be built

• Blank (diluent) injection to show no interference at the retention times of interest.
• System suitability solution or standard, in the number of replicates the method specifies.
• Evaluate every system suitability criterion before sample injections are processed.
• Samples, with bracketing standards at the interval the method or SOP defines, and a closing standard.
• A pre-defined criterion for bracketing standards, for example agreement with the initial standards within 2.0%.

Method validation under ICH Q2(R2)

ICH Q2(R2) “Validation of Analytical Procedures” reached Step 4 in November 2023 and replaced Q2(R1). It was issued together with ICH Q14 “Analytical Procedure Development”. Q2(R2) keeps the familiar performance characteristics but reorganises them, and it extends to newer techniques. For every parameter explained in more depth, see the method validation guide (ICH Q2). The main practical changes for an HPLC analyst are:

• Linearity sits inside “range”. The guideline now speaks of the reportable range and the working range, and of the response (calibration model) within it, with the detection limit and quantitation limit as the lower range limits.
• Robustness is evaluated during development under Q14, and the evidence is carried into the validation package.
• Accuracy and precision may be evaluated together in a combined approach.
• Development data may be used as part of validation if it was generated appropriately.

Which characteristics for which test

Minimum data the guideline asks for

• Specificity. Show no interference from blank, placebo, known impurities and degradation products. For a stability-indicating method this includes forced degradation (acid, base, oxidation, heat, light) and peak purity assessment. See the stability studies guide (ICH Q1A) for how stability data are generated.
• Response (linearity). A minimum of five concentrations across the range. Report the regression equation, correlation coefficient, y-intercept and a residual plot.
• Accuracy. A minimum of nine determinations over at least three concentration levels covering the reportable range, for example three levels with three replicates each. Report percent recovery.
• Repeatability. A minimum of nine determinations covering the reportable range (three levels, three replicates each), or a minimum of six determinations at 100% of the test concentration.
• Intermediate precision. The effect of different days, analysts and equipment within the same laboratory.
• Detection and quantitation limits. By signal-to-noise (3:1 for DL, 10:1 for QL) or from the calibration curve: DL = 3.3σ ÷ S and QL = 10σ ÷ S, where σ is the standard deviation of the response and S is the slope.

Reportable range the guideline recommends

Acceptance criteria

ICH Q2(R2) does not set numerical acceptance criteria. They must be justified in your validation protocol from the intended use and the specification. The figures below are what I most often see in approved protocols for a finished-product assay. Treat them as typical practice, not as guideline requirements.

Validation, verification and transfer are different things

• Validation (ICH Q2(R2), USP <1225>) is for a new or non-compendial method.
• Verification (USP <1226>) shows that a compendial method works for your product in your laboratory. You do not repeat full validation, but you do demonstrate specificity and precision at least, and more where the matrix requires it.
• Transfer (USP <1224>) qualifies a receiving laboratory to run a method validated elsewhere, usually by comparative testing against pre-agreed acceptance criteria.

Calculations with worked examples

Every analyst should be able to reproduce what the CDS reports. If you cannot calculate it by hand, you cannot review it. Carry full precision through each step and round only the final result to the number of decimal places in the specification.

Example 1: system suitability from a chromatogram

Two adjacent peaks: impurity A at 6.20 min (width at half height 0.18 min) and the main peak at 7.10 min (width at half height 0.20 min). For the main peak, the width at 5% height is 0.46 min and the front half-width at 5% height (d) is 0.20 min. Hold-up time is 1.50 min.

• Resolution = 1.18 × (7.10 − 6.20) ÷ (0.18 + 0.20) = 1.18 × 0.90 ÷ 0.38 = 2.8
• Plate number = 5.54 × (7.10 ÷ 0.20)² = 5.54 × 35.5² = 5.54 × 1,260.25 = about 6,980
• Symmetry factor = 0.46 ÷ (2 × 0.20) = 1.15
• Retention factor = (7.10 − 1.50) ÷ 1.50 = 3.7

Example 2: %RSD of replicate standard injections

Five standard injections give areas of 1,254,310; 1,251,872; 1,256,904; 1,249,566 and 1,253,118.

• Mean = 6,265,770 ÷ 5 = 1,253,154
• Deviations from the mean: +1,156; −1,282; +3,750; −3,588; −36
• Sum of squared deviations = 29,917,400
• Standard deviation = √(29,917,400 ÷ 4) = √7,479,350 = 2,734.8
• %RSD = 2,734.8 ÷ 1,253,154 × 100 = 0.22%

Note the divisor of n − 1. Using n gives a smaller, wrong number. The RSD calculator shows each step.

Example 3: assay of a tablet by external standard

Label claim 500 mg per tablet. Average tablet weight (20 tablets) 652.0 mg.

• Standard: 50.2 mg of reference standard (potency 99.6% as is) dissolved to 100 mL; 5 mL diluted to 50 mL.
• Sample: 65.4 mg of tablet powder dissolved to 100 mL; 5 mL diluted to 50 mL.
• Mean standard area 1,254,300; mean sample area 1,248,900.

Standard concentration = 50.2 × 0.996 ÷ 100 × 5 ÷ 50 = 0.0499992 mg/mL.

Sample total dilution volume = 100 × 50 ÷ 5 = 1,000 mL.

Content per tablet = (sample area ÷ standard area) × standard concentration × sample dilution volume ÷ sample weight × average weight

= (1,248,900 ÷ 1,254,300) × 0.0499992 × 1,000 ÷ 65.4 × 652.0

= 0.995695 × 0.0499992 × 1,000 ÷ 65.4 × 652.0 = 496.3 mg per tablet

Percent of label claim = 496.3 ÷ 500 × 100 = 99.3%

Check the basis of the standard potency (as is, anhydrous or dried) against the certificate, and apply a salt-to-base factor if the label claim is expressed as the base. The assay calculator runs the same calculation, and the measurement uncertainty calculator estimates the uncertainty of the result.

Example 4: an impurity against a diluted standard with a relative response factor

Sample solution 1.0 mg/mL of drug. Standard solution 0.001 mg/mL of the drug (equivalent to 0.1%). Standard peak area 15,200. Impurity peak area in the sample 9,880. Relative response factor (RRF) of the impurity 1.3.

Impurity (%) = (impurity area ÷ standard area) × (standard concentration ÷ sample concentration) × (1 ÷ RRF) × 100

= (9,880 ÷ 15,200) × (0.001 ÷ 1.0) × (1 ÷ 1.3) × 100 = 0.65 × 0.001 × 0.7692 × 100 = 0.05%

Check how your monograph defines the factor. Some texts give a correction factor to multiply by, which is the reciprocal of the response factor. Applying it the wrong way round is a common and serious error.

Example 5: detection and quantitation limits from the calibration curve

Residual standard deviation of the regression σ = 1,250 area units. Slope S = 25,000 area units per µg/mL.

• DL = 3.3 × 1,250 ÷ 25,000 = 0.165 µg/mL
• QL = 10 × 1,250 ÷ 25,000 = 0.50 µg/mL

Confirm the calculated QL experimentally by injecting a solution at that concentration and showing acceptable precision and signal-to-noise.

Example 6: scaling a method to a shorter column under USP <621>

Original isocratic method: 250 mm × 4.6 mm, 5 µm, 1.0 mL/min, 20 µL injection. Proposed column: 150 mm × 4.6 mm, 3 µm.

• L/dp original = 250,000 µm ÷ 5 µm = 50,000. L/dp new = 150,000 ÷ 3 = 50,000. No change, so within −25% to +50%.
• Flow rate = 1.0 × (4.6² × 5) ÷ (4.6² × 3) = 1.67 mL/min (a further ±50% is permitted, so 1.0 mL/min may also be used if system suitability is met).
• Injection volume = 20 × (150 × 4.6²) ÷ (250 × 4.6²) = 12 µL

Then run system suitability and document the verification.

Data integrity: ALCOA+ and audit trails

Chromatography is where inspectors look first, because a CDS makes it easy to do the wrong thing and equally easy to find out afterwards. The expectations are set out in the MHRA “GxP Data Integrity Guidance and Definitions” (2018), the FDA guidance “Data Integrity and Compliance With Drug CGMP: Questions and Answers” (2018), PIC/S PI 041-1 (2021) and WHO TRS 1033 Annex 4 (2021), with 21 CFR Part 11 and EU GMP Annex 11 covering electronic records and computerised systems.

Practices that fail inspections

• Trial or “test” injections of samples before the official sequence. Column conditioning is done with a standard or a blank, as written in the SOP, never with the batch sample.
• Testing into compliance: re-injecting or re-preparing until a passing result appears, without an investigation.
• Deleting or overwriting data, or saving results outside the controlled project.
• Unjustified manual integration. Manual integration is allowed only under an SOP that defines when it may be used, requires the reason to be recorded and requires second-person review of before and after chromatograms.
• Disabled audit trails or administrator rights held by analysts who generate data.
• Aborted runs with no documented reason.

Audit trail review

The audit trail records who did what, when and why: method changes, sequence edits, reprocessing, integration changes and deletions. Review of the relevant audit trail entries should be part of the review of each batch record before release, under a written procedure, and periodic system-level reviews should check user rights, time settings and administrator activity. A reviewer who looks only at the printed report has not reviewed the data.

Common failures and where to look first

The table gives the first checks. Each symptom has a full diagnostic path in the HPLC troubleshooting guide.

Change one thing at a time and write down what you changed. Troubleshooting that is not recorded is, from an inspector’s point of view, undocumented manipulation of the system. If the failure affected reportable results, raise a deviation.

Frequently asked questions

What is system suitability in HPLC and why is it required?

System suitability is a set of checks, run before and during an analysis, that shows the complete chromatographic system is working well enough for the method on that day. USP <621> requires it, and sample results are valid only if the system suitability criteria in the monograph or validated method were met.

What is the acceptance limit for tailing factor in HPLC?

The harmonised USP <621> sets a default symmetry factor (tailing factor) of 0.8 to 1.8 for the peak used for quantification, unless the monograph states otherwise. Many monographs and in-house methods specify not more than 2.0. Always apply the limit written in your method.

How many standard injections are needed for system suitability?

Use the number stated in the monograph or validated method. Five or six replicate injections are usual. For drug substance assays, USP <621> gives maximum %RSD values for three to six injections that depend on the upper assay limit; for example, not more than 0.85% for six injections when the assay range is 98.0 to 102.0%.

What is the difference between HPLC and UPLC (UHPLC)?

Both work on the same principle. UHPLC uses columns packed with particles smaller than 2 µm and instruments built for much higher pressure and lower extra-column volume. The result is faster runs, sharper peaks and lower solvent use. “UPLC” is a Waters trade name; “UHPLC” is the general term.

What is the difference between a C8 and a C18 column?

Both are reversed-phase packings bonded to silica. C18 (USP L1) has an 18-carbon chain and retains non-polar compounds more strongly. C8 (USP L7) has an 8-carbon chain, gives less retention and shorter run times for hydrophobic compounds. They are not interchangeable in a pharmacopoeial method.

Why does retention time shift in HPLC?

The common causes are a change in mobile phase composition or pH, unstable column temperature, insufficient equilibration, a leak or pump fault that changes the flow rate, and column ageing. Check the mobile phase preparation and the column oven first, then flow rate and pressure.

Can I change the column dimensions of a pharmacopoeial HPLC method without revalidation?

Yes, within the limits in USP <621>. Column length and particle size may be changed if the ratio of length to particle size stays within −25% to +50% of the original, with flow rate and injection volume scaled accordingly. The stationary phase chemistry must stay the same, system suitability must pass, and the change should be documented with a verification.

What is the difference between ICH Q2(R1) and ICH Q2(R2)?

ICH Q2(R2), finalised in November 2023, replaces Q2(R1). It covers a wider set of analytical techniques, places linearity within the concept of range, allows accuracy and precision to be assessed together, treats robustness as part of method development under ICH Q14, and allows suitable development data to support validation.