Efforts to artificially produce intelligence via association of computing methodologies to cooperatively achieve higher performance system requirements have reached considerable maturity. In many fields, hybrid systems have succeeded to provide theoretical basis and methodological support to build real world applications. For instance, intelligent systems applications in areas such as modeling, control, optimization, classification, forecasting, information search and retrieval, image processing and recognition are common nowadays. The Fifth International Conference on Hybrid Intelligent Systems HIS’05 was a major forum that brought state of the art information to researchers and practitioners that came worldwide. Neural networks, evolutionary computation, fuzzy systems, learning, search, support vector machines, clustering, classification, swarm intelligence, agents, artificial intelligence, to mention a few, were among the techniques used in hybrid systems suggested by authors in their papers. This special issue collects contributions that nurture hybrid intelligent systems growth, emphasizing a balance between new developments and applications. The papers selected for this special issue focus on hybrid intelligent systems in ensembles. Committees or ensembles are receiving more attention from the computational intelligence community in order to improve the performance of pattern recognition systems. They produce a consensus decision that is potentially more accurate than individual models. This strategy is particularly useful when the available committee members are individually efficient and err in different regions of the feature space. After preliminary indications made by conference papers reviewers, anonymous referees did a second refereeing process. The result is a comprehensive and representative set of papers whose content mirror the quality of the conference. The first paper, entitled Evolutionary multiobjective optimization for the design of fuzzy rule-based ensemble classifiers, by Ishibuchi and Nojima, presents a hybrid system, based on evolutionary multiobjective optimization algorithms, applied to the design of fuzzy rule-based ensemble classifiers. The idea is to attain a better performance by creating an ensemble, with high diversity, in a high-dimensional pattern classification problem. To reduce the curse of dimensionality problem, that is, the exponential increase in the total number of possible rules with the increase in the number of attributes, authors proposed a two-stage fuzzy rule selection scheme, where, in the first stage, a pre-specified number of promising fuzzy rules are generated as candidate rules, and in the second stage a genetic rule selection algorithm is applied to the set of candidate rules to construct fuzzy rule-based classifiers with high accuracy and high interpretability. The proposed hybrid system was applied to six high-dimensional data sets, providing interesting results.