We estimate the power spectrum of mass density fluctuations from peculiar v
elocities of galaxies by applying an improved maximum likelihood technique
to the new all-sky SFI catalog. Parametric models are used for the power sp
ectrum and the errors, and the free parameters are determined by assuming G
aussian velocity fields and errors and maximizing the probability of the da
ta given the model. It has been applied to generalized cold dark matter (CD
M) models with and without COBE normalization. The method has been carefull
y tested using artificial SFI catalogs. The most likely distance errors are
found to be similar to the original error estimates in the SFI data. The g
eneral result that is not very sensitive to the prior model used is a relat
ively high amplitude of the power spectrum. For example, at k = 0.1 h Mpc(-
1) we find P(k)Omega(1.2) = (4.4 +/- 1.7) x 10(3)(h(-1) Mpc)(3). An integra
l over the power spectrum yields sigma(8)Omega(0.6) = 0.82 +/- 0.12. Model-
dependent constraints on the cosmological parameters are obtained for famil
ies of CDM models. For example, for COBE-normalized ACDM models (scalar flu
ctuations only), the maximum likelihood result can be approximated by Omega
n(2)h(60)(1.3) = 0.58 +/- 0.11. The formal random errors quoted correspond
to the 90% confidence level. The total uncertainty, including systematic e
rrors associated with nonlinear effects, may be larger by a factor of simil
ar to 2. These results are in agreement with an application of a similar me
thod to other data (Mark III).