The first sign that the storm mattered came not as drama but as data. On August 16, 1992, a tropical wave emerged from Africa and began the long westward journey across the Atlantic. By August 22, the system had become Tropical Depression Three; on August 23, it was named Tropical Storm Andrew. The numbers in the forecast offices began to change faster than the mood on the ground, because the storm was not following the familiar script of weakening and drifting. It strengthened, then weakened, then strengthened again, each turn reminding forecasters that the atmosphere is not a machine but a negotiation.
At the National Hurricane Center in Coral Gables, the warnings were grounded in instruments and models that were good enough to be useful and uncertain enough to be humbling. Andrew’s track shifted southward, and that shift mattered immediately. A few tens of miles changed which counties would bear the brunt. Residents in South Florida had heard hurricane advisories before, but a forecast cone is a strange instrument of fear: it broadens with uncertainty, and people often read its center as a promise rather than a possibility. The storm’s projected path toward the Bahamas and then the Florida Peninsula tightened the area of danger even as its eventual landfall remained difficult to pin down.
That uncertainty had real consequences in the hours when decisions had to be made. Forecasts are not only scientific products; they are public documents, operational guides, and, in practice, the basis for evacuation orders, school closures, and the opening of shelters. In late August 1992, the warnings moved through the familiar channels: weather advisories, television graphics, radio bulletins, emergency management briefings. But each update also carried an implied audit of what South Florida had built, and how well those structures might endure. A hurricane warning is never just about the storm offshore. It is about the exposed inventory inland: roofs, windows, mobile homes, power lines, road capacity, gas supplies, and the limits of time.
The atmosphere itself was becoming more organized. Andrew crossed the Bahamas on August 24 with intensifying winds and a tightening eye, and the storm’s structure improved as it moved over warm water with little shear. That was the critical warning sign that the science could see but the public could not feel. A hurricane is not dangerous only because it is large; it is dangerous because it becomes efficient. The eye clears, the eyewall sharpens, the pressure drops, and the wind field learns how to focus its energy into a moving blade. In meteorological terms, the storm was consolidating. In practical terms, that meant the margin for error was collapsing.
The warning signs were also being read in the architecture of everyday life. By then, South Florida residents had begun the familiar ritual of storm preparation: plywood stacked beside garages, shutters checked, bathtubs filled, batteries bought in bulk. Stores emptied of supplies; gas stations developed lines; emergency managers pushed evacuation advice. But the human problem in any warning phase is that preparation takes time and money, while uncertainty invites delay. Families had to balance the cost of leaving against the possibility that the worst would pass them by. The forecast cone widened and shifted, but the ordinary pressures of work schedules, childcare, fixed incomes, and transportation remained fixed. Some residents would later say they had never imagined a storm could be as violent as the one that came. That gap between warning and comprehension would become one of the defining facts of the disaster.
There were also signals hidden in the built environment, even before the first winds arrived. Contractors knew which houses had been hurried, which roofs had been fastened poorly, which neighborhoods had been built with the expectation that no major storm would expose the shortcuts. Yet ordinary buyers rarely see the structure hidden behind drywall and soffits. A suburban neighborhood can look orderly while its skeleton is compromised. That was the tension inside the warning period: the meteorological system was telling one story, while the construction system had already written another. The danger was not only that Andrew was strengthening; it was that much of what it would strike had never been fully prepared for the test.
The final hours before landfall were marked by the narrowing of options. By the evening of August 23 and into August 24, evacuation orders were being issued for the most exposed areas, especially in the southern parts of Dade County. Families loaded cars with clothing, medicines, photographs, pets, and what little they could fit. For those who stayed, the reasons were often practical rather than defiant: old age, no vehicle, the cost of a motel, the belief that the house had survived other storms, the hope that this one would turn. These were not abstract choices. They were decisions made under clock pressure, with traffic, fuel, and lodging all uncertain, and with the storm’s path still being refined in the evening advisories.
The National Hurricane Center’s advisories became more urgent as Andrew approached the peninsula from the southeast. The storm’s core was compact but extremely intense, a fact that made it easy to underestimate. A small eye can lull an observer into thinking the danger is localized, when in fact the wind around it can be catastrophic. That is the surprising fact that matters most in the warning phase: a hurricane need not be sprawling to be devastating. Andrew’s size concealed its violence. The same compactness that made it easier to draw on a map also made it more deceptive in the field.
At local shelters and in living rooms lit by television weather maps, people watched the same arrows and cones. The warnings were clear enough that no honest account can say South Florida was uninformed. But a warning is not the same as comprehension. Many of the people who would later climb over ruined walls or stand in front of splintered neighborhoods had heard the storm was dangerous. Few had yet grasped how much of the city’s housing stock would fail not at the edge, but at the center. That hidden vulnerability was one of the storm’s crucial secrets. What looked solid from the street could still be vulnerable at its joints, its attachments, and its weakest connection points.
By late evening, the storm was close enough that the horizon itself seemed to have a pulse. The sky lowered, the wind sharpened, and the last normal errands ended. In the hours before dawn, Andrew closed the distance between forecast and impact — and once the first outer bands reached the coast, the line between anticipation and disaster disappeared. What remained was the record: the advisories, the track changes, the evacuation decisions, the preparedness actions taken too late or not at all, and the quiet forensic truth that the warning signs had been present even when their meaning had not fully entered public belief.