Hurricane Andrew crossed Biscayne Bay and made landfall near Fender Point, around SW 280th Street, at 5:05 a.m. on August 24, 1992. Category 5. Sustained winds of 165 miles per hour, though that number wouldn't be settled for years. The storm was compact and fast, which meant the worst of it fell on a corridor roughly twenty miles wide, bounded by Coral Gables to the north, Homestead to the south, the Everglades to the west, and the bay to the east.
Within that corridor, approximately 49,000 homes were destroyed and 108,000 damaged. In Homestead, the hardest-hit community, 1,167 of 1,176 mobile homes were completely destroyed. Ninety-nine percent. The dollar figure eventually settled at $26 billion.
But the number that mattered came later. A nail count.
Daylight
The subdivisions that had filled South Dade's former agricultural land during the 1980s building boom were, in many cases, gone. At Lakes by the Bay, directly in the eyewall's path, the National Weather Service documented "complete destruction." Naranja Lakes, a few miles south, appeared in damage photograph after damage photograph. Homestead Air Force Base, which had stopped routine weather observations before the eyewall arrived because its instruments couldn't survive, was wrecked.
In those first days nobody was asking hard questions. They were pulling people out, counting the missing, trying to get water and ice to a quarter million people whose infrastructure had ceased to exist. But the pattern was visible to anyone walking those streets before the questions were formally asked. Two houses on the same block in the same subdivision, built the same year, under the same building code, by different contractors. One reduced to its slab. The other damaged but upright. A 1945 residence near Homestead, built with 22-inch concrete and coral-rock walls, held together well enough that the people sheltering inside could watch their barometer drop through the eyewall passage while windows shattered around them. They stayed at their instruments, recording pressure readings in a house that predated the South Florida Building Code, while across the road, houses built forty years later under that code were kindling.
Wind doesn't read addresses. Within days, county officials knew something besides wind speed was sorting the destroyed from the standing. The Miami-Dade County Manager's office convened a task force. The county Building Department began commissioning forensic work. The hard questions had arrived.
Reading the rubble
In the weeks after landfall, teams of engineers began walking the damage zone with a purpose different from rescue. They were performing forensics on buildings.
The most systematic effort came from Siddiq Khan and Associates, a forensic engineering firm hired by the Miami-Dade County Building Department and the County Manager's Task Force. Mohammed S. Khan and his colleague Wimal Suaris would become two of the most frequently cited names in the post-Andrew technical literature, and what they did was straightforward and painstaking: they walked seven residential subdivisions, looked at what was broken, and documented whether what was broken had been built the way the code said to build it. The seven subdivisions were Country Walk, American Homes, Hampshire Homes, Deerwood Homes, Lakes by the Bay, Saga Bay, and Naranja Lakes. They produced a ten-volume report titled "Identified Code Violations and Construction Deficiencies."
FEMA sent its own investigators. The American Society of Civil Engineers held a Miami symposium in December 1993, where Ronald A. Cook presented a South Florida overview, Khan and Suaris presented their findings on design and construction deficiencies, and specialists including Edward L. Keith, Gene S. Sanders, and others presented on specific failure modes in sheathing, walls, and truss systems. A Dade County Grand Jury convened.
What they found, collectively, was a story about nails.
The nail-by-nail record
Anybody who's spent time on a construction site knows the difference between a framing nail driven flush into a rafter and one that missed. You can feel it through the gun. The forensic record that emerged from South Dade was specific enough to make a carpenter wince.
Roof sheathing, the plywood panels that form the structural skin of a roof, was supposed to be nailed at six inches on center along panel edges and twelve inches on center in the interior field. Investigators found panels fastened at spacings well beyond those minimums. They found staples where nails were required, staples at wrong spacing and wrong orientation. They found nails that had missed the underlying rafter entirely, driven through plywood into air, holding nothing to nothing. In many of these homes, the plywood sheathing was the only thing bracing the roof trusses laterally. When the sheathing tore away because it wasn't properly fastened, the trusses had nothing keeping them upright. They folded like cards.
The roof-to-wall connection was supposed to be the critical link in the load path, the chain of structural connections that transfers wind force from roof to wall to foundation to ground. Hurricane clips, the small metal straps that tie a roof truss to the top of the wall, were in many cases simply absent. Where they existed, investigators found clips attached only to the top plate of the wall rather than to the wall studs below. Rafters connected to walls by nothing more than toenails, a few nails driven at an angle. In a 90-mile-per-hour wind, toenails might hold. In a 165-mile-per-hour wind, they were a gesture.
The walls told the same story. South Florida's dominant residential wall system was reinforced hollow concrete block masonry. The South Florida Building Code prescribed specific reinforcing: tie beams, tie columns, horizontal reinforcing, tie anchors. FEMA's investigators found poor mortar joints, missing tie beams, missing reinforcing, and missing anchors. Khan and Suaris documented tie-downs placed in unfilled block cells, hooks missing from tie-downs to tie beams and foundations, missing corner bars. In wood-frame construction: improperly constructed corner studs, inadequately nailed connections, missing hurricane straps in stud-to-plate connections, missing sill-plate anchors. Some two-story homes had no direct anchorage between the second-story walls and the first. Metal straps that were supposed to anchor structural members were merely bent under floor joists rather than fastened to them.
Each finding, by itself, was a construction deficiency. Taken across subdivision after subdivision, they described something else. They described buildings that had been inspected, approved, and sold to families as code-compliant homes. Buildings families had signed mortgages on. Buildings families had put their children to sleep in.
What held
The standing buildings were evidence too.
Investigators noted that fully engineered buildings performed well because they received more attention to connections and load paths. No truss failures were cited as a primary cause of building failure when trusses were properly anchored. Masonry walls tended to outperform wood-frame walls because of mass and continuous-system behavior. Homes with shutters and precut plywood over windows showed improved performance. The 1945 coral-rock house near Homestead was an illustration of the whole problem. Somebody had built that house with 22-inch walls because they knew what hurricanes did, and fifty years later it was still standing while code-inspected construction around it was rubble.
The code worked. When someone actually built to it.
The Insurance Institute for Property Loss Reduction estimated that roughly 25 percent of Andrew's insured losses, approximately $4 billion, were attributable to construction that failed to meet code because of poor enforcement and poor workmanship.
That number has been argued about ever since. Engineers who had walked the subdivisions and counted the missing straps and empty block cells considered it conservative, because their violation counts suggested the problem was more pervasive than a quarter of losses implied. Builders and their representatives pushed back, arguing that the estimate unfairly attributed wind damage to construction quality when the storm had exceeded the design assumptions underlying the code. Both sides had evidence. But you could stand in a roofless house in Country Walk and count the staples where nails should have been, and the staples didn't care about the argument.
Twenty-seven years without a test
How did these buildings pass inspection?
The question sounds like an accusation, and it became one in the Grand Jury proceedings. But the honest answer requires you to understand the system these inspectors worked inside, the failures they missed and the conditions that made those failures invisible.
Florida had no statewide building code. More than 400 local jurisdictions and state agencies developed, amended, administered, and enforced their own codes through their own processes. Dade County had its South Florida Building Code, which was by reputation one of the stronger codes in the state. The code text was not the problem. The problem was the distance between what the text required and what got built, and the inspection system that was supposed to close that gap.
During the 1980s, South Dade was one of the fastest-growing areas in the country. Subdivisions were going up on former tomato fields and nursery land at a pace that would have strained any inspection system. The accessible record doesn't provide exact inspector headcount or daily caseload numbers for that period. That absence is itself a kind of evidence. A 2006 University of Florida study for the Florida Building Commission found that even years later, public information about required inspections varied greatly by county, with some counties listing detailed inspection requirements and others listing little or none. The inspection process itself was simple enough: an inspector visited the site, checked the work against the code, and signed a permit card kept on-site. The question was how many sites one inspector could visit in a day, how long they could spend at each one, and what they could realistically see in a framed wall before the drywall went up.
What the record does provide is the Insurance Institute's summary of contributing factors: a long period without a major hurricane, complacency, "lackluster" code enforcement, shortcuts in building practice, and violations that undermined both the code and the building stock.
South Florida hadn't taken a direct hit from a major storm since Betsy in 1965. Twenty-seven years. An entire generation of builders, inspectors, and homeowners had operated in a world where the code's wind-resistance requirements were theoretical. Nobody working in the trades had personal experience of what 165-mile-per-hour winds did to a missed connection. The code said to put the hurricane strap there, and everybody more or less understood why, but understanding why in the abstract and having watched a roof leave a house because the strap wasn't there are different kinds of knowledge.
The argument the rubble couldn't settle
Among the investigators, a debate took shape that would determine everything that followed. Was the problem the code, or the enforcement of the code?
If the code was inadequate, you needed a new code. If enforcement was inadequate, you needed a new enforcement system. If both, the political and financial implications multiplied considerably, because now you were telling every jurisdiction in the state that their codes might be wrong and their inspectors might not be catching the errors.
The Florida Building Commission's investigation came down on the enforcement side: "widespread code violations leading to catastrophic structural failures." University of Florida researchers agreed, finding the problem was "not weakness in the codes themselves" but inability to enforce and comply with multiple codes across a fragmented administrative landscape. The broader technical community, as synthesized by researchers at the Natural Hazards Center, spread responsibility more widely: enforcement, workmanship, code amendments that had weakened requirements over time, and construction shortcuts that had become standard industry practice.
Nobody was operating with certainty. The wind field itself was contested. The National Hurricane Center's preliminary report acknowledged that official surface stations near landfall hadn't captured peak intensity because instruments failed or stations were outside the eyewall. NHC requested data from the public through local media and received more than a hundred quantitative observations, some from people watching barometers and anemometers as their homes came apart around them, recording numbers in conditions that would make most people hide. Some of those observations were rejected. Others were retained because they formed a physically consistent pattern. Andrew was originally classified Category 4 at Florida landfall with estimated 145-mph sustained winds, then upgraded to Category 5 with 165 mph after reanalysis years later. The preliminary report conceded:
"A universally accepted landfall wind speed might be elusive."
If you couldn't specify the exact wind at a specific house, you couldn't say with absolute precision what the wind should have destroyed versus what it shouldn't have. The forensic engineers worked inside that uncertainty. But the patterns they documented didn't require a precise wind speed to read. A hurricane strap that isn't there isn't there whether the wind is 130 or 165. An empty block cell where a reinforcing bar was supposed to be is empty regardless. A nail driven into air holds nothing.
The threshold
The investigations set substantial things in motion. The Grand Jury issued findings. The County Manager's Task Force used SKA's ten-volume report as a basis for revising the South Florida Building Code. The broader question of Florida's fragmented code system entered political discussion. Engineers who had walked the subdivisions of Country Walk and Naranja Lakes began publishing, presenting, and testifying about what they'd found in the rubble.
All of that belongs to the story of what was done with the evidence. The evidence itself was simpler and harder. A nail that missed a rafter. A strap that connected to a top plate and stopped there, an inch short of the stud that would have completed the load path. A block wall with an empty cell where reinforcing was supposed to be. A house that a family had bought believing it was built to code, because an inspector had signed the card, because the system said it was built to code, because nobody had reason to doubt the system. The system had worked fine for twenty-seven years. It had worked fine because nothing had tested it.
The rubble held the truth the functioning system couldn't tell. It took a Category 5 hurricane to perform the audit that the inspection process was supposed to have performed, one nail at a time.
-
Florida's fragmented code legacy: The 1998 Florida Legislature authorized development of a single statewide Florida Building Code after the Building Codes Study Commission recommended reform, a process the Florida Building Commission's own history traces directly to Andrew's exposure of more than 400 jurisdictions enforcing different standards.
-
The $4 billion attribution debate: Jon K. Ayscue's 1996 working paper for the Natural Hazards Center synthesized the competing post-Andrew damage assessments, including the Insurance Institute's estimate that 25% of insured losses stemmed from code violations, and remains one of the most detailed publicly accessible technical accounts of what investigators found.
-
When the wind number changed: NOAA's Hurricane Research Division explains how Andrew was upgraded from Category 4 to Category 5 at Florida landfall after improved understanding of violent eyewall wind fields, a revision that arrived years after the code-versus-enforcement debate had already been framed around the original intensity estimate.
-
Adaptation that outlasts the disaster: The Miami Conservancy District, formed after Dayton's 1913 flood, now describes aging infrastructure, wetter precipitation trends, and an outdated funding model as current risks, a reminder that the governance structures disasters create must themselves be maintained long after the event that justified them.