We present a new ''second generation'' reconstruction algorithm for ir
regular sampling, i.e. for the problem of recovering a band-limited fu
nction from its non-uniformly sampled values, The efficient new method
is a combination of the adaptive weights method which was developed b
y the two first named authors and the method of conjugate gradients fo
r the solution of positive definite linear systems. The choice of ''ad
aptive weights'' can be seen as a simple but very efficient method of
preconditioning. Further substantial acceleration is achieved by utili
zing the Toeplitz-type structure of the system matrix. This new algori
thm can handle problems of much larger dimension and condition number
than have been accessible so far. Furthermore, if some gaps between sa
mples are large, then the algorithm can still be used as a very effici
ent extrapolation method across the gaps.