For any triple of positive integers (rn, N, q), the matrix F(m, N, q), call
ed the (m, N, q)-regular Fourier matrix, is defined. The regular Fourier ma
trices F(m, N, q) are then applied to set up new algorithms for nonuniform
fast Fourier transforms. Numerical results show that the accuracies obtaine
d by our algorithms are much better than previously reported results with t
he same computation complexity. The algorithms require O(N . log(2) N) arit
hmetic operations, where N is the number of data points.