What Worked
Six things did the work. None of them was a ban on its own. The pattern running through all six is that the treaty was built to be changed as evidence arrived.
Binding rules that got stricter over time
The 1987 treaty asked for only a 50% cut by 1998, then was tightened again and again as the evidence grew. It ended up removing about 99% of controlled chemicals. [4] [8]
Paying for the switch
The fund set up in 1990 covered the extra costs poorer countries faced, turning a rich-country deal into a near-global one. It started work in 1991 and was running fully by 1993. [5]
Independent air monitoring
Measuring stations caught the CFC-11 breach from the air itself, without relying on what governments reported. [10] [11]
A science report every four years
A standing panel of scientists reported on a fixed schedule, so the rules could follow the evidence instead of the politics. [8]
Trade limits on countries that stayed out
Staying outside the treaty cost money, so almost everyone joined. [4]
The power to tighten without reopening the treaty
This is how the 1996 deadline change and the later HFC cuts were added, without renegotiating the whole agreement each time. [4] [9]
The fund is the most under-credited item on this list. In 1987 poorer countries used few CFCs and had no reason to give up the cheapest coolant they could buy. Without money on the table the ban would probably have moved production instead of ending it. [5]
What Was Avoided
We know unusually well what did not happen. NASA modelled a world where the treaty was never signed. In it, about two thirds of global ozone is gone by 2065, there are holes over both poles all year round, and harmful UV is up around 650%. Sunburn takes about five minutes. [20]
UV rise in a world with no treaty, by 2065 [20]
650%
Controlled chemicals phased out, 2022 [8]
99%
A warning about reading this page. It is easy to source what worked, because the organisations involved publish their wins. It is harder to source what failed. Read the what failed page next, and treat the length difference between the two as a fact about the record rather than a fact about the world.