This paper presents a survey of multiple classifier systems (MCS) from the perspective of hybrid intelligent systems. MCS combine several classifiers to improve classification performance by leveraging the strengths of individual classifiers. The paper discusses key issues such as diversity and decision fusion methods, and provides an overview of current applications. MCS are particularly effective in handling complex, real-world problems involving ambiguity, uncertainty, and high-dimensional data. The paper also reviews the historical development of MCS, including early works by Chow, Dasarathy, and others, and highlights the importance of diversity in classifier ensembles. It discusses different MCS topologies, including parallel and serial, and explores various methods for ensuring diversity in classifier ensembles, such as diversifying input data, outputs, and models. The paper also covers decision fusion strategies, including class label fusion, support function fusion, and trainable fusers. It addresses the challenge of concept drift, where statistical dependencies between features and classification may change over time, and discusses how MCS can adapt to such changes. The paper concludes with a review of current applications of MCS in remote sensing, computer security, financial risk assessment, fraud detection, and medical diagnosis. The growing popularity of MCS is reflected in the increasing number of publications and their widespread use in various domains.This paper presents a survey of multiple classifier systems (MCS) from the perspective of hybrid intelligent systems. MCS combine several classifiers to improve classification performance by leveraging the strengths of individual classifiers. The paper discusses key issues such as diversity and decision fusion methods, and provides an overview of current applications. MCS are particularly effective in handling complex, real-world problems involving ambiguity, uncertainty, and high-dimensional data. The paper also reviews the historical development of MCS, including early works by Chow, Dasarathy, and others, and highlights the importance of diversity in classifier ensembles. It discusses different MCS topologies, including parallel and serial, and explores various methods for ensuring diversity in classifier ensembles, such as diversifying input data, outputs, and models. The paper also covers decision fusion strategies, including class label fusion, support function fusion, and trainable fusers. It addresses the challenge of concept drift, where statistical dependencies between features and classification may change over time, and discusses how MCS can adapt to such changes. The paper concludes with a review of current applications of MCS in remote sensing, computer security, financial risk assessment, fraud detection, and medical diagnosis. The growing popularity of MCS is reflected in the increasing number of publications and their widespread use in various domains.