A new technique of identifying the flow regimes of air/water two-phase flow in a vertical pipe is proposed. This technique is based on analyzing the statistical characteristics of the static and differential pressure signals by an optimum modeling method. The major concept of the optimum modeling method is to fit the two-phase flow pressure noise by autoregressive moving average (ARMA) models with an optimization technique. The results show that it is possible to identify the flow patterns from a set of “flow regime indices,” such as dynamic signature, order of dominant dynamics mode, and order of ARMA model. A computer code based on these indices has been built on an IBM-PC/XT microcomputer to perform two-phase flow pattern identification. The success probability of this code is ∼85% on the data base collected from our experimental work. The experimental data points are also indicated in a Taitel flow map and excellent matching has been shown, except for some points around the flow regime transition boundaries. These discrepancies are due to the subjective categorization of the flow regimes.