Analysis of 24 years of lunar laser ranging data is used to test the p
rinciple of equivalence, geodetic precession, the PPN parameters beta
and gamma, and G over dot/G. Recent data can be fitted with a rms scat
ter of 3 cm. (a) Using the Nordtvedt effect to test the principle of e
quivalence, it is found that the Moon and Earth accelerate alike in th
e Sun's field. The relative accelerations match to within 5x10(-13). T
his limit, combined with an independent determination of gamma from pl
anetary time delay, gives beta. Including the uncertainty due to compo
sitional differences, the parameter beta differs from unity by no more
than 0.0014; and, if the weak equivalence principle is satisfied, the
difference is no more than 0.0006. (b) Geodetic precession matches it
s expected 19.2 marc sec/yr rate within 0.7%. This corresponds to a 1%
test of gamma. (c) Apart from the Nordtvedt effect, beta and gamma ca
n be tested from their influence on the lunar orbit. It is argued theo
retically that the linear combination 0.8 beta+1.4 gamma can be tested
at the 1% level of accuracy. For solutions using numerically derived
partial derivatives, higher sensitivity is found. Both beta and gamma
match the values of general relativity to within 0.005, and the linear
combination beta+gamma matches to within 0.003, but caution is advise
d due to the lack of theoretical understanding of these sensitivities.
(d) No evidence for a changing gravitational constant is found, with
\G over dot/G\less than or equal to 8X10(-12)/yr. There is significant
sensitivity to G over dot/G through solar perturbations on the lunar
orbit.