Study design

How to choose a sample size for an animal study

Enough animals to answer the question, and no more: the five numbers you need.

The five inputs

1. The smallest effect worth detecting: for example a 30% fall in liver triglycerides. Base it on biological importance, not on what would give a convenient n.
2. Variability: the standard deviation, from a pilot study, your own previous work or published data.
3. Significance level (α): usually 0.05.
4. Power: usually 80% or 90%; the chance of detecting the effect if it is real.
5. Expected losses: animals that may reach humane end points or be excluded.

A worked example

Control mean 100, expected change 30%, SD 25, α 0.05 two-sided, power 80%: about 12 animals per group, or 14 allowing for 10% loss. Try your own numbers in the free calculator.

Good practice

• Decide the primary outcome first; the sample size is calculated for that outcome.
• More complex designs (several groups, repeated measures, factorial) need the matching calculation.
• Report how the sample size was decided; ethics committees and ARRIVE 2.0 both ask for it.
• The 3Rs: a well-powered study is part of Reduction, because underpowered studies waste animals on inconclusive results.

Published 2026 by Dr. Ambuj Shahi. This guide is general information for researchers; follow your institution’s and journal’s requirements.