An estimate of the largest loss a risk can reasonably suffer under adverse but plausible conditions.
Probable Maximum Loss is an estimate of the largest loss an insured risk is likely to suffer when protective measures function normally, under an adverse but realistic scenario. It differs from Maximum Possible Loss, which instead assumes the failure of all protections and represents a theoretical upper bound rarely reached. PML is a central tool for underwriting and accumulation management, as it allows capacity to be sized, retentions to be set and reinsurance to be structured without reasoning on the absolute worst case, which is economically untenable. Its determination rests on assumptions, the scope of the loss, the effectiveness of barriers and the chosen scenario, which makes it a conventional figure rather than an objective measure. This is precisely its weakness, since two experts may arrive at materially different PMLs for the same risk depending on scenario assumptions. In cyber, the notion is delicate to transpose, because the maximum loss depends on shifting digital dependencies and dynamic threats, making PML less stable than for a localized industrial risk.
For a plant fitted with automatic sprinkler systems, the fire PML might be estimated at 30 percent of the insured value, against a maximum possible loss of 100 percent should all protections fail.
perte maximale probable, sinistre maximum probable, SMP