Chance of Rain Calculator (South Africa)

How often does it actually rain where you live?

Most rainfall pages give you an average, in millimetres, for a month. That is the wrong number for planning. Nobody cancels a wedding because of millimetres. What you want to know is how many days of that month are actually wet, and whether the month you are looking at is meaningfully better than the one before it. This tool answers that for 29 South African towns using 47 years of daily rainfall.

South Africa makes this unusually worth checking, because the country runs on two opposite rainfall calendars. The Highveld, the Lowveld, the Free State and most of KwaZulu-Natal get their rain in summer, so Johannesburg swings from about 2% of days wet in July to about 45% in December. The south western Cape does the reverse, with Cape Town at about 5% in February and about 27% in June. The southern Cape coast does neither and simply drizzles all year, which is why George sits between about 16% and about 23% in every single month of the year.

Planning outdoor work is a different question

Painting, waterproofing, plastering and concrete do not care whether one particular day is wet. They care whether you can get several dry days in a row. That is a different statistic and it is the one nobody publishes for South Africa, so the second tab works it out: pick the job, see the odds of an uninterrupted run by starting month.

It also settles an argument. Search for the best time to paint a house in South Africa and you will find national advice saying winter, and other national advice saying summer. Both are half right. Starting a five day job in Johannesburg in July works out about 93% of the time against about 11% in December, so winter wins on the Highveld. In Cape Town the same job is about 79% in February against about 29% in June, so winter is the worst thing you could pick. Neither piece of advice is wrong, they are just about different halves of the country.

What this tool will not do

It will not tell you which Saturday in November to choose. That was the original plan and testing killed it: splitting the record in half and asking each half which day of a month is driest, the halves agreed 4% of the time, which is chance. Ranking days is reading noise, so this tool ranks months and says so wherever it might be tempting to read more into a chart than the data holds.

When is the rainy season in South Africa?

There is no single answer, because the country runs two opposite calendars and a third region that barely has a season at all. On the Highveld, the Lowveld, the Free State and most of KwaZulu-Natal the rain arrives with the first thunderstorms around October and runs to about March, so Johannesburg goes from roughly 2% of days wet in July to about 45% in December. The south western Cape is the mirror image, wettest in June at about 27% of days and driest in midsummer. The southern Cape coast, from Mossel Bay through George to Plettenberg Bay, gets rain in every month of the year and has no meaningful dry season at all.

For the fuller picture, ocean currents, temperature and frost as well as rainfall, see our South Africa Climate Guide.

Which rainfall region is my town in?

South Africa has three, and which one you are in decides everything about when to plan an outdoor job. The split below is measured rather than drawn by eye: it is the share of a town's annual rainfall that falls in the winter half of the year, April to September. Under 40% is a summer rainfall town, over 60% is a winter rainfall town, and in between is a town with no strongly dry half.

Summer rainfall, 17 of 29 towns

Rain arrives with the thunderstorms around October and runs to about March, and winter is the dry half. This is most of the country: the Highveld, the Lowveld, the Free State, the Karoo interior and almost all of KwaZulu-Natal. Johannesburg runs from about 2% of July days wet to about 45% in December.

Bethlehem, Bloemfontein, De Aar, Durban, East London, Howick, Johannesburg, Kimberley, Mahikeng, Makhanda, Nelspruit, Newcastle, Pietermaritzburg, Polokwane, Pretoria, Rustenburg, Upington.

Rain in every season, 8 of 29 towns

Mostly the southern and eastern Cape coast, where rain is spread through the year and no month is reliably dry. Calvinia in the Karoo lands here too, on the winter edge of the group. George sits between about 16% and 23% of days wet in every month of the year, which is why the tool refuses to name a best month for these towns rather than picking one out of noise.

Calvinia (close to the winter-rainfall group), George, Gqeberha (close to the summer-rainfall group), Knysna, Mossel Bay, Oudtshoorn, Plettenberg Bay, St Francis Bay.

Winter rainfall, 4 of 29 towns

The south western Cape, the mirror image of the rest of the country. Most of the rain falls between about April and September and midsummer is the dry half, so Cape Town is wettest in June at about 27% of days and driest in February at about 5%.

Cape Town, Malmesbury, Paarl, Stellenbosch.

If your own town is not listed, pick the one in your group that is most like yours in landscape, a coastal town for a coastal town and an inland one for an inland one, rather than the closest on a map. Thirty kilometres over a mountain range can be a different rainfall world, which is the same reason the town list is short in the first place. Two towns are marked above as sitting close to a boundary, and for those the honest reading is that they behave a little like both neighbours.

Driest and wettest month for every town

Rain days, not millimetres. A month can carry a big total in two downpours and still be mostly dry, which is why the millimetre figures published elsewhere are a poor guide to whether your day will be rained on. Where several months are too close to separate, this table says so instead of picking one. There is one table per rainfall region, so the towns listed alongside yours are the ones whose weather actually resembles it.

Summer rainfall towns, 17 of 29

Rain roughly October to March, dry winters.

Towns with rain in every season, 8 of 29

No strongly dry half of the year, the southern and eastern Cape coast.

Winter rainfall towns, 4 of 29

Rain roughly April to September, dry summers, the south western Cape.

Rain days a year counts days with at least about a millimetre, the same standard the South African Weather Service uses. The last column is the month to begin a five day outdoor job for the best chance of finishing without a wet day.

Where these numbers come from

Every figure on this page is computed from 47 years of daily rainfall, 1979 to 2025, taken from the ERA5 reanalysis through the Open-Meteo historical archive. Reanalysis is not a set of rain gauge readings. It is a physical model of the atmosphere replayed over the past and constrained by the observations of the day, which is what makes a consistent daily record reaching back to 1979 possible at all, and it is also the source of the limits below.

The model runs on a grid roughly 25 km across, so it reproduces South African annual totals well but spreads them across too many days, reading light drizzle where a gauge would record nothing. Two corrections are applied in that order. First, each town's annual total is scaled onto its own published rainfall normal, from the South African Weather Service or NOAA. Where a town's published normal covered a period too old to compare against a modern model window, it was dropped rather than corrected onto, because the difference between the two periods would have been imported as if it were model bias. Second, a single national threshold of 2.15 mm defines a wet day, which is what reproduces the real number of rain days despite sounding higher than the roughly one millimetre a weather office uses.

Those two steps are checked against different things on purpose. The scaling is fitted on annual millimetres; the result is then validated against published rain-day counts, which were not used to fit it. 12 of the 29 towns have such a count published and come out with a mean error of about 8%. The rest have an annual normal only, and say so on screen. A town with no published normal at all is not listed, because there would be no way to know whether its answer was right.

The grid is also why the town list is short rather than exhaustive. Sampling one metro at two different points inside its own boundary moved its annual total by 47%, so a town without its own published normal to correct against is genuinely unknown rather than probably fine. The underlying record is refreshed once a year, in March, after the archive settles. Adding a year moves the typical answer by about a fifth of a percentage point, well inside the margin of error already shown on every figure.

Common questions

Related tools