Four named storms crossed Florida in six weeks in 2004. Charley struck the southwest coast on August 13 as a Category 4. Frances made landfall on the central east coast on September 5. Ivan crossed the western Panhandle on September 16. Jeanne followed Frances's approximate path on September 26. Wilma came through South Florida the next year, on October 24, 2005.
These were the first sustained test of the Florida Building Code, which had taken effect in March 2002. Some of the houses built under it had been standing for two and a half years. Sorting out what the storms said about the code's performance took years, and the evidence arrived unevenly: engineering observations within months, statewide economic evidence more than a decade later.
What Held
The Florida Building Commission funded a survey of 195 houses in Charlotte, St. Lucie, and Escambia counties: site-built single-family homes put up between 1994 and 2004, selected by construction era and mapped wind exposure rather than by known damage. The researchers split the sample into pre-code groups, from the Standard Building Code era of 1994 to 2001, and a Florida Building Code group built from 2002 to 2004.
The structural findings were clean. None of the surveyed houses lost roof sheathing, walls, or roof-to-wall connections, including houses exposed to estimated gusts as high as 150 mph. Field teams had seen those failures in houses built before 1994, but the older stock was excluded from the statistical sample, so the comparison runs between post-1994 code eras rather than between the new code and the weakest construction on the ground.
Wind-rated garage doors in the sample held. Protected windows failed significantly less often. In Charley's highest mapped gust zones, close to 30 percent of code-era houses with shingle roofs lost no shingles at all, while every sampled house built between 1994 and 1998 lost at least some.
FEMA's assessment teams reached a compatible conclusion. Structural systems designed and built under current requirements generally performed as expected where the storm delivered design-level winds, and most damage to main structural systems occurred in older buildings.
Where Rain Got In
The same evidence that documented structural improvement documented a persistent failure of the building envelope, meaning the components that keep weather out: roof coverings, soffits, windows, doors, rooftop equipment. Across construction eras, across storms, across the state, these kept letting wind-driven rain inside.
The 195-home survey found no clear improvement in water penetration by era. Water came in through windows, exterior walls, doors, and ceilings in houses built before and after the code took effect. The researchers identified soffit failure in code-era construction as a significant path for water entry and flagged double-entry-door integrity as a continuing concern.
Storm characteristics mattered to the comparison. Charlotte County showed less water penetration at comparable wind speeds partly because Charley moved through quickly, while Frances and Jeanne exposed St. Lucie County to longer rainfall and to a second storm weeks after the first. But the researchers were direct about the bottom line: it was not clear that the Florida Building Code improved water-penetration outcomes.
FEMA attributed the pattern to a combination of design requirements, component performance, attachment methods, installation quality, maintenance, and workmanship. A house could keep its structural system intact, roof tied to walls, sheathing attached, and still take on enough water through a failed soffit to put the family somewhere else for months while the ceilings and drywall came out.
Miami-Dade and Broward counties were operating under the High Velocity Hurricane Zone provisions, the stricter requirements Miami-Dade had fought to keep inside the statewide code during the 1999 drafting, including the enhanced sheathing attachment, mandatory sheathing inspections, and more rigorous glazing standards. Wilma was the first major hurricane to test whether that differentiated standard performed differently.
Miami-Dade's post-Wilma assessment reported minimal structural damage, strong performance from the sheathing attachment and the mandatory sheathing inspections the county had insisted on keeping, and little damage to recent glazing. Mortar-set hip and ridge tiles, first-course mechanically attached tiles, and poorly installed screen enclosures stayed vulnerable. The county attributed many roof losses to installation deficiencies and inadequate workmanship, and proposed code changes eliminating mortar-set attachment and requiring post-installation testing. Where glazing was breached, inspectors found interior water intrusion and internal pressurization, though the county reported that the damaged glazing it examined was not in buildings constructed under the latest code.
The stricter provisions did appear to produce better outcomes in the areas they addressed: sheathing, connections, glazing. Installation quality remained the part the code could not reach on its own, however stringent the page was.
County by County
The field survey documented something the 1998 reform's architects would have recognized. What actually got installed varied by jurisdiction. Roof-to-wall connections at every truss appeared in 95 percent of sampled Charlotte County houses, 92 percent in St. Lucie, 83 percent in Escambia. Among 1994–1998 houses in St. Lucie County, the figure was 65 percent.
Those numbers show variation in construction. They do not by themselves establish that inspector staffing or enforcement intensity caused it. A later commission survey of sixty-three jurisdictions found respondents rating funding sufficiency at 3.5 out of 5 and its importance at 4.8. They reported shortages of field personnel, limited expertise in small jurisdictions, inconsistent interpretations, and trouble retaining qualified inspectors.
The strongest statistical evidence that enforcement intensity mattered came from a peer-reviewed analysis using Insurance Services Office grading data. After accounting for the statewide code's effect, jurisdictions with more intensive local administration and enforcement saw an estimated additional 15 to 25 percent reduction in losses. The statewide code dominated the result; local implementation measurably amplified it.
The 1998 compromise had kept local enforcement as one of its five foundations. The 2004–2005 storms showed how much the foundation's strength varied from one building department to the next, and that the variation reached the houses. A family in Charlotte County and a family in Escambia County, living in homes built to the same code, were not facing the same odds that the code had been followed while their house went up.
When the Evidence Arrived
The code's performance did not become visible all at once. It came in pieces, through different institutions, on different clocks.
The Florida Building Commission received preliminary storm-performance information in October 2004 and convened a researchers' workshop that December. FEMA published its three-storm summary in March 2005. University of Florida researchers released preliminary comparative results from the field survey in June 2005. The full 195-home report followed in March 2006.
Those were engineering observations from a relatively small sample. They could show that structural systems held and that water kept getting in. They could not say what the code was worth statewide.
A 2018 analysis using ten years of insured-loss data estimated the Florida Building Code reduced windstorm losses by as much as 72 percent, with a benefit-cost ratio of roughly six to one and an estimated eight-year payback period.
That number arrived in a 2018 Land Economics analysis drawing on ten years of insured-loss data: windstorm losses reduced by as much as 72 percent, roughly six dollars of avoided loss for every dollar of added construction cost, an estimated eight-year payback. A companion paper examining wind characteristics across seven hurricanes found statistically significant post-code loss reductions after controlling for wind speed, duration, and directional change.
Thirteen years passed between the storms and that figure. In the interval, the code's defenders had field observations and a 195-house survey, and its critics had the water damage, the uneven enforcement, and the cost of compliance. Both were arguing from partial views of the same thing.
What the Evidence Establishes
The peer-reviewed record supports several claims with reasonable confidence. The Florida Building Code measurably reduced wind losses compared to pre-code construction. Structural systems performed well where design-level winds arrived. More intensive local enforcement amplified the effect. The economic return substantially exceeded what compliance cost.
The same record does not support treating the code as finished work. Water intrusion persisted across construction eras. Envelope components stayed vulnerable even when the structure underneath held. Enforcement capacity varied across jurisdictions, and the variation reached outcomes. The researchers behind the 195-home survey cautioned that aging and code effects could not always be cleanly separated, which kept them from putting a sharper number on how much of the improvement the code itself produced.
The reform took ten years to build, and it changed what happened to houses in a storm. It did not close the distance between a requirement written in Tallahassee and a nail driven correctly into a truss in Escambia County. That was local work, and it depended on which department reviewed the plans, which budget paid for inspectors, and who showed up the day the roof sheathing went on.
