East African savannah straddling the border, where the two rainy seasons — not temperature — decide when to go. The serengeti and masai mara barely has a temperature season: daytime highs move only about 3 °C across the whole year, from 25 °C in May to 28 °C in October. What changes instead is rain.
Measured against a city trip, the conditions hold up May to August. June is the most consistent single month: highs near 25 °C, nights around 14 °C, and roughly 14 mm of rain across the month.
The binding constraint here is rainfall and temperature — that is what rules months out, not any composite judgement. Rainfall is sharply seasonal: December averages 148 mm against 12 mm in July, so the wet season is the real planning constraint rather than heat or cold. February is the weakest month on these criteria.
January, September, November sit just outside the window — part of the month works and part does not. These are usually the interesting weeks: conditions are close to the peak-season answer while prices and crowds are not. The map lets you check them fortnight by fortnight rather than trusting a monthly average.
Month by month
| Month | High °C | Low °C | Precip mm | Wet days |
|---|---|---|---|---|
| January | 26 | 15 | 109 | 14.3 |
| February | 27 | 15 | 82 | 10.9 |
| March | 27 | 16 | 120 | 13.5 |
| April | 26 | 16 | 126 | 15.8 |
| May | 25 | 15 | 73 | 9.0 |
| June | 25 | 14 | 14 | 3.8 |
| July | 25 | 14 | 12 | 2.0 |
| August | 26 | 14 | 27 | 3.5 |
| September | 27 | 15 | 23 | 4.4 |
| October | 28 | 16 | 46 | 6.0 |
| November | 27 | 16 | 107 | 12.5 |
| December | 26 | 15 | 148 | 15.6 |
Ticked rows are months where every hard requirement for a city trip is met across at least half the month.
Open Serengeti & Masai Mara on the map →Nearby regions
Figures are 30-year climatology sampled at Seronera, Ndutu, Lobo, Talek and Sekenani, averaged across those locations, from ERA5-Land reanalysis fitted with a harmonic seasonal cycle and a trend term, evaluated for 2026. Sampling named towns rather than averaging the whole region avoids blending valley and mountain conditions into a figure true of neither. Cells are about 0.08° wide, from ERA5-Land's 0.1° grid, which is itself driven by a 0.25° (~28 km) atmospheric reanalysis — so individual valleys, sheltered bays and single summits are not resolved. In steep terrain read these as conditions for the wider area rather than for one village. They describe typical conditions, not a forecast. Read how the climatology is built.