When is it going to stop raining? For any specific location, no honest forecast extends beyond about 10 days, and useful rain-timing detail runs out after roughly 48 to 72 hours. Rain ends when the lifting mechanism that produced it weakens, the moisture supply is cut off, or the storm system moves away. A single shower typically lasts 20 to 60 minutes, while stratiform rain can persist for 12 to 36 hours.
The Life Cycle of a Rain Cloud
Rain begins with rising air. The sun heats the ground unevenly, warm air becomes buoyant, and it climbs. As that parcel rises, pressure drops and it expands and cools at roughly 5.4°F per 1,000 feet in unsaturated air. Cool enough air reaches its dew point, water vapor condenses onto microscopic particles called cloud condensation nuclei, and a cloud forms. Every raindrop starts as one of these condensed droplets, typically 10 to 20 micrometers across.
Droplets do not fall at that size because rising air holds them up and because they are too small to beat drag. They grow by collision and coalescence, sweeping up smaller droplets as they drift, and in cold clouds by the Bergeron process, where ice crystals grow at the expense of supercooled liquid. A droplet must reach roughly 0.02 inches (about 0.5 millimeters) in diameter to survive the trip to the ground. Below that, it evaporates into subsaturated air beneath the cloud base, a process called virga.
Rain stops when one of three things fails. The updraft feeding the cloud shuts off. The moisture supply is exhausted or diverted. Or the entire system advects away on the wind. A shower is a self-terminating machine: its own downdraft, cooled by evaporating rain, spreads out and cuts off the warm inflow that sustains it.
That single mechanism explains the counterintuitive detail professionals rely on. Heavy rain often ends faster than light rain. Intense convection consumes its fuel quickly and collapses, while weak stratiform rain can drizzle for a full day because nothing shuts it down.
How Long Rain Lasts, by Storm Type
Storm type sets the ceiling on duration more than any other variable. Use this table as a rough planning guide; the ranges are typical, not guaranteed.
| Rain type | Typical duration | Spatial scale | What ends it |
|---|---|---|---|
| Single air-mass shower | 20–60 minutes | 1–5 miles | Downdraft cuts off updraft |
| Multicell cluster | 1–3 hours | 10–50 miles | Cold pool undercuts inflow |
| Supercell thunderstorm | 1–4 hours | 10–30 miles | Outflow dominates or storm weakens |
| Squall line | 2–6 hours | 100–300 miles long | Passage of the gust front |
| Stratiform rain shield | 12–36 hours | 100–500 miles | Parent cyclone moves on |
| Orographic rain | 6–24 hours | Localized to windward slopes | Wind shifts or moisture depletes |
| Tropical cyclone rainband | 6–24 hours per band | Hundreds of miles | Eyewall passage, then storm exit |
The pattern matters. A pop-up afternoon thunderstorm in summer is a 30-minute event. A winter frontal system can soak a region for a day and a half. Knowing which regime you are in is more useful than watching a single radar frame.
Quantified Context: Rates, Totals, and Forecast Skill
Rainfall intensity spans four orders of magnitude. Drizzle falls at less than 0.01 inches per hour. Light rain runs 0.01 to 0.1 inches per hour. Moderate rain is 0.1 to 0.3 inches per hour, and heavy rain exceeds 0.3 inches per hour. Torrential downpours can hit 2 to 4 inches per hour. A 30-minute heavy shower at 1 inch per hour delivers just 0.5 inches, while a 12-hour moderate rain at 0.15 inches per hour delivers 1.8 inches. Duration and intensity trade off.
Forecast skill degrades predictably with lead time. Day-one probability of precipitation forecasts are well calibrated. By day five, timing errors of several hours are routine. Beyond day 10, the atmosphere's chaotic sensitivity makes deterministic rain timing impossible in principle, not just in practice. Ensemble forecasting exists precisely because single runs diverge.
Climate normals give the base rate. In the contiguous United States, it rains or snows on roughly 8 to 12 percent of days in most locations, varying from under 5 percent in the desert Southwest to over 15 percent in parts of the Pacific Northwest and Gulf Coast. That base rate anchors your prior. A forecast of rain in Seattle in November is a very different bet than one in Phoenix in June.
Why Your Phone Says 40 Percent and It Is Pouring
The probability of precipitation number is widely misread. It is not the chance it rains at your exact coordinates. It is the chance that measurable rain (at least 0.01 inches) falls somewhere in the forecast area during the forecast period. A 40 percent chance can mean a 40 percent chance of a brief shower anywhere in a county, or a near-certain light rain over 40 percent of the area. Both produce the same number.
Timing granularity compounds the confusion. An hourly forecast showing a 60 percent chance at 3 p.m. describes a one-hour window. The daily number aggregates overlapping windows, so a day with six wet hours at 30 percent each can post a misleadingly low daily figure. Radar loops and short-term nowcasts, which extrapolate storm motion for 0 to 6 hours, beat daily percentages for the question of when rain stops.
Radar nowcasts lose accuracy past 60 to 90 minutes as storms grow and decay. Hourly precipitation probability is often a smoothed model output, not a fresh observation. "Scattered" means coverage under about 50 percent; "numerous" means over 50 percent. A dark sky downwind does not mean rain is arriving; virga and upper clouds are common. Rain can begin before radar shows it if the beam overshoots low cloud or is blocked by terrain.
Edge Cases That Break the Rules
Orographic enhancement is the most underappreciated duration extender. When moist air is forced up a mountain slope, it cools and rains continuously as long as the wind direction holds. Windward slopes can accumulate 2 to 3 times the lowland total, and the rain can persist for a full day even with no storm overhead.
Training storms are the flood producers. Individual cells each last less than an hour, but if they repeatedly form and move over the same ground, the effective rain duration at one location can run 6 to 12 hours. Flooding is governed by accumulation rate against soil infiltration, not by how long any one cloud lives.
Cutoff lows are the slow-motion exception. A low-pressure system detached from the jet stream can stall for days, spinning moisture ashore like a stuck sprinkler. These systems produce some of the longest continuous rain episodes outside the tropics, sometimes 48 to 72 hours with breaks.
Tropical systems add a distinct timeline. An outer rainband may pass in 1 to 2 hours. Then comes a lull, then the eyewall, which can last 1 to 3 hours with the heaviest rain, then the eye's calm, then a mirrored sequence on exit. Total exposure can span 12 to 24 hours, and the lull is not the end.
Practical Actions: Reading the Signals Yourself
You can answer the when-will-it-stop question more accurately than a generic app with a few disciplined checks.
Pull a radar loop, not a still image. Storm motion over 30 to 60 minutes tells you the arrival time far better than any percentage. Check the 0 to 6 hour nowcast for your coordinates; treat anything beyond 6 hours as a trend, not a promise. Look at the cloud type overhead. Towering cumulus means a short, self-limiting shower. Flat, uniform stratus means steady drizzle that may last hours. Note the wind shift. A veering or backing wind often precedes a frontal passage that will clear or sustain rain. Compare observed rain rates to your soil. Over 1 inch per hour on saturated ground raises flood risk within the hour. Treat a forecast lull with suspicion if a cutoff low or tropical system is nearby. Lulls between bands can last 30 to 90 minutes.
Plain version: follow the storms on radar, watch the sky, and trust nothing past a few hours.
What No Forecast Can Tell You
The theoretical ceiling on rain prediction is not technological. It is the butterfly effect operating in a fluid. Two model runs differing by one part in a million in initial conditions diverge into different weather within about two weeks. That limit will not yield to more computing power; it is built into the physics.
Local terrain adds irreducible uncertainty. A hill 500 feet tall can anchor a shower for an hour while the valley a mile away stays dry. Models with 2-mile grid spacing cannot resolve that. Convective initiation, the moment a cloud decides to become a thunderstorm, remains a research frontier with genuine skill gaps even at day one.
Climate change adds a statistical shift rather than a per-event answer. Warmer air holds more water vapor, roughly 4 percent more per 1.8°F of warming, which tends to raise rainfall intensity more than duration. Heavier bursts and longer dry spells can coexist. A specific storm's stop time remains a nowcast question.
For any individual rain event, the honest answer is a range tied to radar, sky, and system type, refreshed every hour. The atmosphere is a chaotic system, and its stopwatch resets with every new cloud.