Case file
- What happened: IKEA's Malm line of chest of drawers tipped forward when unanchored, crushing at least eight children over a span of several years.
- Scale: 17 million units recalled across the United States and Canada in June 2016.
- Root cause: Inherent instability under foreseeable consumer misuse – multiple drawers opened simultaneously, unit not anchored to a wall, weight loaded unevenly.
- The bill: Largest furniture recall in CPSC history, undisclosed wrongful-death settlements, permanent brand association with child fatalities.
17Munits recalled in the US and Canada
8+child deaths linked to Malm tip-overs
2016year of the full product recall
The situation
The Malm was one of IKEA's highest-volume furniture lines. Affordable, flat-packed, sold by the million. Lightweight particleboard construction, stability contingent on the consumer anchoring the unit to a wall with a bracket provided in the box. The voluntary US standard ASTM F2057 sets stability requirements for clothing storage units above a certain height and weight threshold. The Malm dressers covered by the recall did not reliably meet this standard when tested under conditions approximating actual household use — including scenarios with multiple drawers open at once.How it unfolded
The first child fatalities were reported years before the 2016 recall. IKEA's initial response — public warnings, free anchoring kits, awareness campaigns — treated the deaths as a consumer-education gap. The product was safe if used as instructed. The problem lay with consumers who did not follow instructions. That framing survived multiple fatalities. It was not until mounting CPSC pressure, media coverage, and a growing death toll forced the issue that IKEA recalled 17 million Malm and other model dressers in June 2016. Additional deaths were reported even after the recall — units still in homes, still unanchored.Root-cause anatomy
Two failure layers, stacked. Technical first: when drawers are opened on a freestanding chest, the centre of gravity shifts forward. Open multiple drawers at once and the shift becomes dramatic. Add the weight of clothing — or a child pulling on an open drawer — and the unit tips. This is first-year engineering statics. The forces are calculable during design. The organisational layer: the quality system validated the product against intended use — one drawer at a time, unit properly anchored — rather than against reasonably foreseeable misuse. In PFMEA terms, the failure mode "tip-over due to simultaneous drawer opening while unanchored" was either not identified, or was rated with severity and occurrence scores that did not reflect how people actually interact with affordable furniture in homes with small children.Where the quality system failed
The PFMEA broke. That is where this case was lost. A failure modes and effects analysis that examines only what happens when the consumer behaves perfectly is not an FMEA. It is a fiction. The methodology requires you to ask what happens when the consumer does everything wrong: opens every drawer at once, never reads the manual, leaves the anchor kit sealed in its bag, loads the top drawer with winter sweaters.When the only barrier between your product and a fatality is a printed leaflet, you have already accepted the fatality.The anchoring instruction is an administrative control. It depends on the consumer performing a specific action, with tools, on a wall that may or may not accommodate a bracket. No verification after sale. No audit. The defect is latent and invisible until a toddler reaches for a drawer. In any PFMEA I have signed off, a safety-critical risk whose sole mitigation is "consumer will follow instructions" does not pass review. It goes back to engineering for an inherent-design solution.
What would have caught it
None of this is exotic. It is standard practice in safety-critical manufacturing. A PFMEA that enumerates foreseeable-misuse scenarios explicitly — simultaneous drawer openings, unanchored use, top-heavy loading, carpeted floors, child interaction patterns. Stability testing beyond the minimum standard: tip-tests with all drawers open and weighted, performed unanchored, as a mandatory design-validation gate. Engineering controls instead of warnings. Wider base footprint, lower centre of gravity, counterweighting in the lower drawers, or a drawer interlock mechanism that physically prevents more than one drawer opening at a time. An APQP design-review gate that rejects any safety-critical failure mode mitigated solely by consumer behaviour. And post-market change control that triggers redesign — not a louder warning — when field data shows consumers are not anchoring.My take
I have spent two decades in automotive and aerospace, where foreseeable misuse is not optional in a PFMEA. It is a required input. In AS9100 and IATF 16949 environments, the standard demands it. When I sit in a design review and an engineer proposes "operator will follow SOP" as the sole control for a critical-to-safety characteristic, I stop the meeting. That is not a control. That is a wish dressed up in procedural language. The Malm case, from my chair, is a textbook gap between compliance-driven quality and engineering-driven quality. A team can pass every audit, hold every certificate, and still ship a product that kills someone — if the FMEA only asks the questions the product can survive. The questions that matter are the ones where the answer is ugly. I have seen this pattern on the shop floor: a process that runs perfectly when the operator follows every step in sequence, and produces scrap when they skip one. The answer is not a bigger sign above the workstation. It is a poka-yoke that makes the skip impossible or harmless. The same principle scales from the production line to the living room. If your dresser is lethal without a wall anchor, the anchor is not your solution. Your design is your problem.What this means on your floor
- Design for the worst-case user, not the ideal one. If your FMEA only covers intended use, you have no FMEA.
- Any safety risk mitigated solely by consumer action — a label, a manual, a warning — is an uncontrolled risk. Reject it at design review.
- Post-market data showing that consumers ignore your instructions is not a consumer-education problem. It is a design feedback signal that demands engineering action.
- The hierarchy of controls is non-negotiable: elimination, engineering controls, then administrative measures. A warning label is the last resort, never the first.
