What caused the Hyatt Regency walkway collapse?

NBS concluded that the most probable cause was insufficient load capacity of the box beam–hanger rod connections. Physical evidence strongly indicated initiation at the east end of the fourth-floor walkway's middle box beam.

Did dancing cause the Hyatt Regency collapse?

No. NBS concluded that dynamic effects from walking or dancing were not significant in comparison with the static loads. The fourth-floor connections had inadequate reserve: NBS estimated their mean ultimate capacities below the estimated dead + upper-bound live-load demand at collapse.

Would the original continuous hanger-rod detail have satisfied the building code?

It was safer than the as-built load path in one important respect, but it still did not satisfy the Kansas City Building Code/AISC connection requirements. NBS estimated that the original box-beam connection would nevertheless have resisted the loads actually present on the night of the collapse.

Why did the revised hanger detail double the fourth-floor connection load?

With one continuous rod, the lower walkway load passed through the rod rather than through the fourth-floor box-beam connection. With two rod segments, the lower rod terminated at the fourth-floor box beam. The fourth-floor connection therefore had to transfer both the fourth-floor walkway load and the force from the lower walkway into the upper rod.

What is the main engineering lesson from the Hyatt Regency collapse?

Treat critical connection details and load-path changes as engineered design work; define delegated-design responsibilities explicitly; require calculations with shop-drawing revisions; independently check nonredundant connections; and never let schedule pressure replace structural verification.

Hyatt Regency Connection Design Failure

01 // Interactive failure replay

Select a stage or run the sequence.

Hyatt Regency atrium — load-path reconstructionThe second- and fourth-floor walkways are aligned vertically; the third-floor walkway was offset and independently suspended.

Geometry and collapse motion are schematic. Load values, connection configuration and initiation location follow the NBS investigation where stated.

Structural / process state

Replay encoding

Roof-to-walkway rods

Lower hanger load path

Critical connection / failure

Checked / controlled condition

Key NBS findings

Clear walkway width8.25 ft / 2.51 m

Code live load100 psf / 4.79 kPa

Max 4F design connection action40.7 kip / 181 kN

Max estimated 4F load at collapse21.4 kip / 95 kN

Average 4F ultimate≈18.6 kip / 83 kN

Dynamic dancing effectNot significant

Engineering event console — latest stage first

02 // Connection load-path explorer

See why the shop-drawing change doubled the fourth-floor nut reaction.

Connection arrangement

Illustrative one-walkway connection transfer, P — educational statics model

20.3 kip

Upper rod tension40.6 kip

4F nut / box-beam transfer20.3 kip

Relative connection demand1.00 × P

Meaninglower walkway bypasses 4F beam

Simplified P / 2P model result20.3 kip at 4F nut

NBS maximum as-built design action40.7 kip

Statics boundary. The slider uses a simple symmetric notation, P, to explain the load-path change. NBS's detailed analysis used actual span geometry, dead loads and live-load distribution. Under nominal dead + code live load, NBS found a maximum fourth-floor connection action of 40.7 kip and a maximum second-floor connection action of 20.3 kip. These are results of the detailed NBS load model, not an exact symmetric P/2P pair; the simplified explorer therefore illustrates the topology change rather than reproducing every NBS connection reaction.

03 // What physically failed?

The rod did not simply “snap.” The box-beam connection deformed around the nut and washer.

As-built fourth-floor connections — NBS benchmark quantities

Average estimated ultimate

≈18.6 kip

Maximum estimated 4F load at collapse

21.4 kip

Maximum 4F dead + code live action

40.7 kip

Approx. code-expected ultimate*

≈68 kip

These bars are benchmark quantities, not a demand/capacity pair at one identical connection. The 18.6 kip value is the NBS average estimated ultimate capacity across the six as-built fourth-floor connections; individual means ranged 18.2–19.3 kip. The 21.4 kip value is the maximum estimated load at connection 8UE, while physical evidence indicated probable initiation at 9UE. NBS concluded that the estimated dead + upper-bound live load exceeded the estimated mean ultimate capacity at every fourth-floor connection. *For the code comparison, ≈68 kip is 1.67 × 40.7 kip, reflecting the NBS discussion of the ultimate capacity expected for a connection designed to the applicable AISC/Kansas City code basis. This is not a separate measured load.

Connection failure sequence

The nut and washer bear against the underside of the welded box beam. The concentrated load acts near the joined channel flanges. This three-step graphic is schematic: NBS documented coupled flange/weld separation, deformation and slip-stick washer/nut pull-through rather than one universal microscopic sequence at every connection.

04 // Cause-and-effect chain

Immediate connection failure and engineering-process failure were inseparable.

01 / INCOMPLETE CONNECTION BASISThe contract drawings showed a hanger detail, but the box-beam connection itself did not satisfy applicable AISC/Kansas City provisions even in the original configuration.

02 / LOAD PATH CHANGEDThe shop-drawing change split one continuous rod into two segments with a 4 in offset at the fourth floor, making the upper connection carry both walkway transfers.

03 / NO ADEQUATE RECALCULATIONThe revised shop drawing passed through the engineering review process without a documented calculation demonstrating that the changed box-beam, washer, weld and rod system could carry the new demand.

04 / CONNECTION DEFORMSAt a load well below code design conditions, the critical fourth-floor box-beam connection deformed and the hanger-rod nut/washer pulled through.

05 / PROGRESSIVE COLLAPSELoss of one fourth-floor support transferred load to the remaining connections, which had little reserve; the fourth-floor walkway and the second-floor walkway suspended from it fell.

05 // Engineering figures

Documented loads and the design-review sequence.

Fourth-floor connection — NBS benchmark comparison

Average estimated ultimateMaximum estimated collapse loadMaximum code design action

View data and interpretation

Quantity	Value	Meaning
Average estimated as-built ultimate	≈18.6 kip	NBS average estimated ultimate capacity across the six fourth-floor box beam–hanger rod connections; individual mean estimates ranged 18.2–19.3 kip. This is the estimate for the as-built connections; NBS separately reports a mean ultimate capacity of about 20.5 kip (91 kN) from laboratory tests of a single-rod connection, which is the figure most often quoted in secondary sources.
Estimated maximum at collapse	21.4 kip	NBS upper-bound estimate at the most heavily loaded fourth-floor connection (8UE); NBS identified 9UE as the probable initiation location from physical evidence.
Nominal dead + code live load	40.7 kip	NBS maximum fourth-floor design action using 100 psf live load.

Benchmark comparison only: the 18.6 kip bar is an average capacity across six connections, while 21.4 kip and 40.7 kip are maximum connection actions. NBS nevertheless found that estimated collapse loading exceeded the estimated mean ultimate capacity at every fourth-floor connection. Dancing was not needed to explain the failure.

Process timeline — where control barriers failed

Design / fabricationWarning / review opportunityFailure

View timeline

Date	Event
Dec 1978	Havens Steel contracted for atrium steel fabrication/erection.
Feb 1979	Shop drawings changed from one continuous hanger rod to two rod segments; drawings returned with the engineering review stamp.
Oct 1979	Part of atrium roof collapsed during construction; subsequent reviews created a major opportunity for broader verification.
Jul 1980	Hotel opened.
17 Jul 1981	Second- and fourth-floor walkways collapsed at approximately 7:05 p.m.

Communications surrounding the detail change were disputed. The important engineering fact is that the revised load path appeared on review-stamped shop drawings without an adequate structural verification of the connection.

06 // What actually happened?

The immediate structural problem was inadequate capacity at the fourth-floor box beam–hanger rod connections. NBS concluded that the most probable initiation point was the east end of the fourth-floor walkway's middle box beam [1]. Once that support was lost, redistribution into the other under-strength connections made progressive failure of the suspended system likely.

The shop-drawing change mattered because it changed the load path

In the contract-drawing arrangement, one continuous rod ran from the atrium roof through the fourth-floor connection to the second-floor walkway. The rod above the fourth floor carried force associated with both walkways, but the fourth-floor box-beam connection transferred only the fourth-floor walkway load into that rod. In the as-built arrangement, a separate lower rod hung the second-floor walkway from the fourth-floor box beam, so the fourth-floor connection had to transfer both walkway loads into the upper rod.

But the original detail was not code-compliant either

This distinction is important. NBS found that even the original continuous-rod box-beam detail did not satisfy the applicable Kansas City Building Code/AISC provisions. Its estimated ultimate capacity was still well short of the code-expected connection capacity. However, NBS also concluded that the original arrangement would have resisted the loads estimated to have been present on the night of the collapse.

Dancing did not cause the collapse

NBS specifically concluded that dynamic loads from walking or dancing were not significant relative to the static loads. The maximum estimated fourth-floor connection load at collapse was about 21.4 kip at 8UE—about 53% of the 40.7 kip maximum nominal dead + code live-load connection action calculated for design. NBS separately identified 9UE as the probable initiation location from deformation evidence [1].

Failure was not a materials-quality problem

NBS found that neither workmanship quality nor the structural materials played a significant role in initiating the collapse. The problem was fundamentally one of structural configuration, connection capacity and design/verification process.

07 // Key event and structural data

Parameter	Value / finding
Hotel opened	July 1980
Collapse	17 Jul 1981, ≈7:05 p.m.
Atrium plan	≈117 × 145 ft
Atrium height	≈50 ft
Suspended walkways	2F, 3F and 4F
Collapsed walkways	2F + 4F
3F walkway	Offset; independently suspended
Clear width, 2F/4F	8.25 ft / 2.51 m
Concrete deck basis	3¼ in; 110 pcf in NBS load model
Code live load used by NBS	100 psf / 4.79 kPa
Maximum 4F design connection action	40.7 kip / 181 kN
Maximum 2F design connection action	20.3 kip / 90 kN
Estimated max 4F load at collapse	21.4 kip / 95 kN
Estimated max lower-rod load	11.5 kip / 51 kN
Offset between upper/lower rod segments at 4F	4 in / 102 mm
Human fatalities	114 ultimately reported by ASCE

08 // Original detail vs as-built detail

Same general components, fundamentally different force transfer.

Contract drawing concept — continuous hanger rod

4F connection demand≈P

Upper rod force≈2P

Code statusstill deficient

As built — two hanger-rod segments

4F connection demand≈2P

Load-path change2F hangs from 4F

Effect~2× connection transfer

09 // Engineering lessons learned

The lasting lesson is not merely “check the arithmetic.” It is to control structural responsibility when information changes hands.

Connection details are structural designIf a detail changes the load path, force transfer, stiffness or failure mode, it must be treated as an engineered design decision—not an incidental drafting adjustment.

Shop-drawing approval must be substantiveA review stamp is not a clerical act. Critical connections need a traceable check against design loads, detailing provisions and the engineer’s load-path intent.

Delegation does not erase responsibilityWhen design tasks are delegated to fabricators or specialists, responsibility boundaries, required calculations, submittals and acceptance criteria must be explicit.

Check the whole load path after every changeThe rods themselves were only part of the system. Nut, washer, box-beam flanges, welds, webs and eccentric load transfer all mattered.

Warnings should trigger broader verificationThe 1979 atrium roof collapse created a major opportunity to reassess structural design and quality-control processes before opening.

Public safety outranks schedule pressureFast-track delivery increases the need for disciplined interfaces, revision control and independent checks—not less.

Defense in depth: controls that should stop this class of failure

CalculationExplicit connection design package

Define design actions, load combinations, connection resistance, eccentricity, local flange/web checks, rods, welds, washers and stiffeners.

Change controlLoad-path change review

Any shop-drawing revision that changes support topology should trigger a documented structural recalculation before approval.

Independent checkSecond-person verification

Critical suspended elements and nonredundant connections should receive a check independent of the original designer/detailer.

Interface controlClear delegated-design responsibility

Contract documents should state who designs each connection, what information is provided, what must be sealed and how the engineer of record reviews it.

10 // Historical chronology

Date	Event
Dec 1978	Havens Steel Company contracted to fabricate and erect atrium steel.
Feb 1979	Shop drawings changed the fourth-floor hanger arrangement from one continuous rod to two segments with a 4 in offset. Ethics-case chronology records receipt on 16 February and return with the engineering review stamp on 26 February [5].
14 Oct 1979	Part of the atrium roof collapsed during construction.
Oct–Nov 1979	Investigations and structural assurances followed; the episode represented a major verification opportunity before completion.
Jul 1980	Hyatt Regency Kansas City opened.
17 Jul 1981, ≈7:05 p.m.	Second- and fourth-floor walkways collapsed during a crowded atrium event.
21 Jul 1981	NBS investigators arrived in Kansas City.
May 1982	NBS Building Science Series 143 published the federal investigation findings.
1984	Missouri professional disciplinary proceedings were filed against engineers and the engineering firm involved.

11 // Sources and technical basis

This page uses the National Bureau of Standards investigation as the primary source for the structural mechanics, load values, connection configuration and probable initiation location. ASCE and engineering-ethics sources are used for professional-practice context.

Primary-report section map

For auditability, the principal numerical claims on this page trace to NBS BSS 143: walkway geometry and hanger arrangement (chapter 3); estimated occupancy and collapse loading (chapter 4 and §9.4); material/connection tests (chapters 6–8); code/load analysis (§9.3); probable collapse sequence and hanger-detail effect (§§10.4–10.5); and final findings (chapter 11).

Important casualty-count note

Modern ASCE accounts report 114 fatalities. The NBS/NIST historical web material reflects earlier counts from the investigation period (113 deaths and 186–188 injuries). This page uses 114 as the ultimately reported fatality total and avoids presenting a single definitive injury number.

Visualization boundary

The atrium animation and P/2P load-path explorer are explanatory models. They do not reproduce the full NBS finite/detail analysis, exact crowd distribution, transient collapse kinematics or every connection location.

12 // Frequently asked questions

Common misconceptions and practical lessons.