Acceptance tests for surveillance radars are often formulated in terms
of the cumulative probability of detection (P) at a specified range.
Statistical methods for determining point and interval estimates of P
from a small number of flight tests are discussed. Three types of esti
mators, frequency, Bayesian, and maximum likelihood, are studied both
analytically and by Monte Carlo simulation. Extensions to more complic
ated acceptance tests, involving track formation probabilities, are di
scussed.