Organisations and sources Β· Multi-agency food-security classification system

Integrated Food Security Phase Classification (IPC)

What it is

Developed from 2004 in Somalia; now a global partnership of UN agencies, NGOs and governments ⚠.

Analysis teams of partner agencies in each country, working to common protocols; every famine classification is reviewed by an independent Famine Review Committee.

Who funds it

Donor governments and partner agencies ⚠.

What it wants

In its own words, 'the primary mechanism the international community uses to analyse data and arrive at a conclusion of whether Famine is happening or projected to occur', with protocols 'specifically designed to provide strong early warning signals'.

Its record

  • Its famine definition has three parts that must occur together: 20% of households facing an extreme lack of food, 30% of young children acutely malnourished by weight-for-height (or '~15%' by arm circumference, MUAC, 'when WHZ data is not available'), and more than 2 non-trauma deaths per 10,000 people per day. Its fact sheet adds that the MUAC route is assessed case by case, 'to determine if a MUAC rate at or above 15% signals the same level of concern as a 30% acute malnutrition rate based on WHZ'.
  • Its definitions 'refer strictly to the observed conditions and do not imply causality, intent, or alignment with any legal definitions' β€” a famine classification is not a finding about who caused it.
  • Famines it has classified: Somalia 2011, South Sudan 2017 and 2020, Sudan 2024 and 2025, and Gaza 2025 (see 'Israel is starving Gaza β€” famine has been confirmed.', including how its Gaza page changed after publication).

Citation cluster

Its findings become inputs for UN agencies, NGOs and courts; it is the upstream source, and its own data come from partner agencies' surveys and screenings in the area.

How to read a claim sourced from here

Read the classification with its evidence level ('famine' or 'famine with reasonable evidence'), which thresholds were measured and which inferred, and the dates of the data used. A projection is a warning, not a count.

Where it comes up

Sources