In this paper we point out the drawbacks of conventional target fluctu
ation models used in radar target modeling. It is usually difficult fa
r us to statistically model st real target by a conventional target mo
del which has an analytical probability density function (pdf) express
ion, because there are very few parameters which can be used to approx
imate in conventional target models the pdf of the radar cross section
(RCS) of a real target. We suggest a new method of statistical modeli
ng, where the first nth central moment of the RCS data Ear real target
s, combining with the Legendre orthogonal polynomials, are: used to re
construct the pdf of the RCS of the target, The relationship between t
he coefficients of the Legendre polynomials and the central moments of
RCS are deduced mathematically. Through a practical computing example
, the error-of-fit is shown as a function of the orders of Legendre co
efficients. By comparing the errors-of-fit caused by both the new mode
l and the conventional models, we conclude that the new nonparametric
method for statistical modeling of radar targets is superior, for it m
akes the statistical modeling of radar target easier and more exact.