What is a calibration curve?

A calibration curve is a graphical relationship between a known independent variable (such as concentration) and a measured dependent variable (such as absorbance or signal intensity). You prepare a series of standards with known concentrations, measure their signals, then fit a straight line through the data using least-squares linear regression. Once you have the equation y = mx + b, you can measure the signal of an unknown sample and rearrange the equation to find its concentration: x = (y - b) / m.

What does R² mean for a calibration curve?

R² (the coefficient of determination) measures how well the straight line fits your data. A value of 1.0 means a perfect fit, while 0 means no linear relationship. For analytical calibration curves, R² should typically be 0.999 or higher to be considered acceptable. Values below 0.99 suggest poor linearity, possible outliers in your standards, reagent problems, or that your signal-concentration relationship is not truly linear over the range you are using.

How many standard points do I need for a calibration curve?

Most analytical methods require a minimum of five standard points (including a blank or zero standard) to establish a reliable calibration curve. More points improve the statistical confidence of the line. Standards should bracket the expected concentration range of your unknowns. Including a blank (zero concentration) anchors the line at the origin and helps identify any baseline signal. Regulatory methods (such as EPA or ISO methods) often specify the minimum number of standards required.

Calibration Curve Calculator

This calculator builds a calibration curve, also called a standard curve, from a set of laboratory standards with known concentrations and their measured signal, then fits the best straight line through the data using least-squares linear regression. It suits analytical chemistry work such as spectrophotometry and chromatography, where you need to convert a raw instrument signal into an actual concentration. You enter at least three, ideally five or more, paired data points of known concentration (x) and measured signal (y), add labels and units for each axis, and can add or remove rows as needed. The calculator returns the fitted equation in the form y = mx + b, the slope and intercept, the R² (coefficient of determination) and R (Pearson correlation) values showing how well the line fits, the mean x and y, a plain-language linearity assessment, and a table of fitted values and residuals for every standard so you can spot outliers. Enter the measured signal from an unknown sample and the tool back-calculates its concentration using x = (y − b) / m. Aim for an R² of 0.999 or higher, include a blank standard, and avoid extrapolating beyond the range of your standards. This tool is for general analytical and educational use, so check your results against any method-specific acceptance criteria before reporting them.

Advertise on this page

Reviewed June 2026 Standard least-squares linear regression (ISO 8466-1). Suitable for spectrophotometry, chromatography, and general analytical chemistry.

1. Standard Data Points

Enter the known concentration (x) and measured signal (y) for each standard. Minimum 3 points required.

#	Concentration (x)	Signal (y)

2. Labels and Unknown

X-axis label (e.g. Concentration)

X-axis unit (e.g. mg/L)

Y-axis label (e.g. Absorbance)

Y-axis unit (e.g. AU, leave blank if dimensionless)

Unknown sample signal (y) to back-calculate x

Calibration Results

Slope (m)

Signal per unit x

Intercept (b)

Signal at x = 0

R²
-
Goodness of fit (1.000 = perfect)

Data Points (n)

Valid standards used

Regression Statistics

Slope (m)-

Intercept (b)-

R² (coefficient of determination)-

R (Pearson correlation coefficient)-

Mean x-

Mean y-

Linearity assessment-

Predicted Values (Fitted)

#	x (known)	y (measured)	y (fitted)	Residual

How a Calibration Curve Works

A calibration curve (also called a standard curve) establishes the mathematical relationship between a known quantity and an instrument's measured response. You prepare a series of solutions with known concentrations (the standards), measure each one with the instrument, then fit a straight line through the resulting data points using least-squares linear regression.

Once the line is established, you measure the signal from an unknown sample and solve for its concentration using the rearranged equation: x = (y − b) / m, where m is the slope and b is the y-intercept.

The Least-Squares Method

Least-squares linear regression finds the line y = mx + b that minimises the sum of the squared vertical distances (residuals) between each data point and the line. For n data points (x₁, y₁), (x₂, y₂), ... , (xn, yn):

Parameter	Formula
Slope (m)	m = [n∑xy − (∑x)(∑y)] / [n∑x² − (∑x)²]
Intercept (b)	b = (∑y − m∑x) / n = ȳ − mx̄
R²	R² = 1 − SSᵣᵉᵣ / SSᵗᵒᵗ
Back-calculate x	x = (y − b) / m

Worked Example

Using the default data (5 standards: 0, 2, 4, 6, 8 mg/L with absorbances 0.003, 0.101, 0.198, 0.302, 0.398):

Slope (m) = 0.04955
Intercept (b) = 0.002200
Equation: y = 0.04955x + 0.002200
R² = 0.99987
For unknown signal y = 0.312: x = (0.312 − 0.002200) / 0.04955 = 6.25 mg/L

The R² of 0.99987 indicates an excellent linear fit, as expected for a well-prepared set of spectrophotometric standards.

Interpreting R²

R² Value	Interpretation
0.9999 or above	Excellent. Suitable for most analytical methods.
0.999 to 0.9999	Good. Acceptable for most routine analyses.
0.99 to 0.999	Marginal. Review your standards and technique.
Below 0.99	Poor. Do not use. Investigate outliers or non-linearity.

Practical Tips

Always include a blank (zero concentration standard) to account for any baseline signal.
Standards should bracket the range of expected unknown concentrations. Do not extrapolate beyond your highest standard.
Use at least five standards. More points improve the confidence of the fit.
Prepare standards from a fresh stock solution, not from the same stock used to prepare unknowns.
Re-run standards if R² falls below 0.999 or if the curve shifts between sessions.
Check residuals: large residuals for specific points may indicate outliers or reagent problems.

Related Calculators

Maths and Stats Calculators: full hub for statistics and maths tools.
Linear Regression Calculator: general least-squares regression for any two-variable dataset.
Correlation Coefficient Calculator: Pearson and Spearman correlation.
Standard Deviation Calculator: measure variability in your replicate measurements.
Average Calculator: mean, median, and mode for your data.
Drake Equation Calculator: civilisations in the Milky Way.
Drake Equation for Love Calculator: estimate Potential Partners NZ.

Method: Least-squares linear regression per ISO 8466-1:1986 (Water quality: calibration and evaluation of analytical methods). R² computed as 1 minus the ratio of the residual sum of squares to the total sum of squares. Back-calculation assumes the calibration range is not exceeded.

This calculator is for educational and general analytical purposes. Verify your calibration against any method-specific acceptance criteria before reporting results. Do not extrapolate outside your calibrated range.