The acoustical signal generated by a helicopter, as well as many other signals, e.g., a voiced speech signal, can be described as periodic or almost periodic. When these signals are observed in the presence of other additive wide-band signals, it becomes interesting to separate these two kinds of signals. In this paper we specifically present an iterative method for separating helicopter signals from wide-band interference modeled as an autoregressive (AR) process. The suggested algorithms can also be used for other harmonic signals. Experimental results of our algorithm using real helicopter signals are presented to demonstrate the practical applicability of the method.