Home Library Glossary Debris & Space Safety Pc
🛡️ Debris & Space Safety

Probability of Collision (Pc)

Also known as: Pc, Collision Probability, Collision Risk

Quick answer

Probability of collision (Pc) is the computed chance that two space objects will actually hit during a predicted close approach, combining their miss distance, uncertainty ellipsoids and physical sizes. Operators typically consider avoidance manoeuvres when Pc rises above about 1-in-10,000.

📘 Full definition✓ Reviewed 2026-09-07
Probability of collision is the number that turns a close approach into a decision. A predicted miss distance alone is nearly meaningless — 400 metres is terrifying or trivial depending on how well both orbits are known — so conjunction assessment instead computes Pc: the probability that, given each object's covariance, the true separation at the time of closest approach will be smaller than the combined size of the two objects (their hard-body radius). Mathematically it integrates the combined position uncertainty over a disc the size of the merged objects in the encounter plane. The resulting figures look tiny but carry real weight: common decision thresholds are around 10⁻⁴ (1-in-10,000) for planning a collision-avoidance manoeuvre, with operators tracking the trend across successive CDMs rather than any single value. Pc behaves counterintuitively: it usually rises as tracking improves and covariance shrinks around a genuinely close geometry, and in the opposite regime — very poor data — it can be misleadingly small because the probability mass is smeared over a huge volume, the well-known "dilution" effect. Read properly, Pc is less a verdict than a live conversation between geometry and data quality.
Manoeuvre threshold
~10⁻⁴ (1-in-10,000)
varies by operator and asset
Inputs
Miss + covariance + size
all three, not miss distance alone
Watch the
Trend across CDMs
single values mislead
Gotcha
Dilution regime
bad data can fake a low Pc

Understanding Pc

Why Pc rises before it falls

Days before an event, covariances are large: the probability mass is spread across a wide volume and Pc is modest almost regardless of geometry. As fresh tracking arrives, the ellipsoids shrink. If the refined orbits separate, Pc collapses towards zero and everyone stands down. If they stay tight, the same shrinkage concentrates probability onto the collision cross-section and Pc climbs — sometimes by orders of magnitude in a day. That is why operators never celebrate an early low Pc: the informative numbers are the late ones, computed when uncertainty is smallest, and why the manoeuvre decision is typically held until a day or two before TCA — late enough for good data, early enough to act on it.

What the thresholds mean in practice

PcReadingTypical response
< 10⁻⁷Background riskNo action, routine monitoring
10⁻⁵–10⁻⁷ElevatedWatch the CDM series, refine tracking
> 10⁻⁴ActionablePlan avoidance manoeuvre, coordinate with the other operator
> 10⁻²CriticalRare; act — comparable to the riskiest recorded conjunctions
See it live Watch predicted close approaches between tracked objects on the live conjunction feed — each one is a Pc calculation waiting to resolve. Open the conjunction feed →
📖 Learn More

Frequently Asked Questions

Because miss distance without uncertainty is noise. A predicted 200 m pass with metre-level ephemerides is a controlled non-event; a predicted 2 km pass between two poorly tracked debris objects may carry far more real risk. Pc folds the quality of knowledge into the number, which is exactly what a decision-maker needs.
For a single event it sounds small, but fleets accumulate exposure: a constellation facing thousands of conjunctions a year at that level would expect regular collisions. Thresholds balance the real cost of manoeuvres — propellant, service interruption, downstream conjunctions on the new orbit — against the catastrophic cost of a hit.
When covariance is very large relative to the encounter, the computed Pc becomes small simply because probability is spread thinly — worse data produces a more reassuring number. Practitioners treat low Pc with huge uncertainty as "unknown risk", not "low risk", and respond by seeking better tracking rather than standing down.

Sources & References

Definitions are reviewed against primary sources. Last reviewed: 2026-09-07.