The magnetic anisotropy of europium iron garnet is caused by the combined effect of the crystalline field and anisotropic exchange, of which the latter proves to be the more important. Most of the parameters required for the analysis are derived in part A from the available experimental results on nuclear magnetic resonance, spectroscopy, magnetization and Mössbauer technique. The exchange parameters
G
m
n
,
n
≽ 4 (i. e. fourth and higher degree harmonics) cannot be obtained in this manner. When their values are set equal to zero the magnetic anisotropy constant
K
1
(or rather
K
1
+ ⅓
K
2
) coincides in sign and order of magnitude with the experimental result. From the present study, it becomes apparent that the value of
K
1
is very sensitive to small changes in the parameters
G
m
4
. Consequently the calculations are repeated for different values of the
G
m
4
and it is then possible to satisfy simultaneously the experimental values of
K
1
+ ⅓
K
2
and the available results on the ferromagnetic resonance of Eu i. g. (to be discussed in another paper). The formulae used to derive from the experimental results the parameters required in the calculation are presented in part B, together with a critical discussion of those used to describe the nuclear magnetic resonance of Eu in Eu i. g. The only apparent inconsistency in the theory relates to the quadrupolar substructure in the n. m. r. when averaged over the three principal local axes. It is half as large as predicted theoretically and the discrepancy cannot be blamed on higher order harmonics or uncertainty in the parameters. A possible explanation could be unexpectedly large biquadratic superexchange.