So maybe convolve(x,y) should take care if x or y is of prime size ? Or users could at least be warned that this implementation of "Fast Convolution" is not able to handle x of prime size in acceptable time. Anyway, I don't understand why R does not make use of a fast&free algorithm like the FFTW implementation, which is able to handle x of prime sizes in acceptable time. "Sizes with small prime factors are best, but FFTW uses O(N log N) algorithms even for prime sizes." Ralf.