Flow distribution during gas–liquid–liquid upflow through a vertical pipe is investigated. The optical probe technique has been adopted for an objective identification of flow patterns. The probability density function (PDF) analysis of the probe signals has been used to identify the range of existence of the different patterns. Dispersed and slug flow have been identified from the nature of the PDF, which is bimodal for slug flow and unimodal for dispersed flow. The water continuous, oil continuous, and emulsion type flow distributions are distinguished on the basis of the PDF moments. The method is particularly useful at high flow rates where visualization techniques fail. Based on this, a flow pattern detection algorithm has been presented. Two different representations of flow pattern maps have been suggested for gas–liquid–liquid three phase flow. © 2014 American Institute of Chemical Engineers AIChE J, 60: 3362–3375, 2014