This paper presents eight design guidelines for multiple view systems in information visualization. These guidelines are based on a workshop discussion and the authors' experience in designing and implementing such systems. The guidelines are divided into two sets: one for deciding when multiple views are desirable and another for using multiple views effectively. The first set of guidelines includes diversity, complementarity, parsimony, and decomposition. Diversity suggests using multiple views when there is a variety of attributes, models, user profiles, levels of abstraction, or genres. Complementarity recommends using multiple views when they reveal correlations or disparities. Parsimony advises using multiple views minimally, as they introduce complexity and cognitive overhead. Decomposition suggests partitioning complex data into multiple views to manage information and provide insight into interactions among dimensions. The second set of guidelines includes space/time resource optimization, self-evidence, consistency, and attention management. Space/time resource optimization balances the costs and benefits of presenting multiple views. Self-evidence uses perceptual cues to make relationships among views more apparent. Consistency ensures that the interfaces and states of multiple views are consistent. Attention management uses perceptual techniques to focus the user's attention on the right view at the right time. The guidelines aim to help designers avoid unnecessary complexity and minimize the costs of using multiple view systems. They also address trade-offs between different design considerations and provide a framework for evaluating and improving multiple view systems. The authors conclude that these guidelines will be useful for heuristic walkthroughs of both designs and fully implemented systems.This paper presents eight design guidelines for multiple view systems in information visualization. These guidelines are based on a workshop discussion and the authors' experience in designing and implementing such systems. The guidelines are divided into two sets: one for deciding when multiple views are desirable and another for using multiple views effectively. The first set of guidelines includes diversity, complementarity, parsimony, and decomposition. Diversity suggests using multiple views when there is a variety of attributes, models, user profiles, levels of abstraction, or genres. Complementarity recommends using multiple views when they reveal correlations or disparities. Parsimony advises using multiple views minimally, as they introduce complexity and cognitive overhead. Decomposition suggests partitioning complex data into multiple views to manage information and provide insight into interactions among dimensions. The second set of guidelines includes space/time resource optimization, self-evidence, consistency, and attention management. Space/time resource optimization balances the costs and benefits of presenting multiple views. Self-evidence uses perceptual cues to make relationships among views more apparent. Consistency ensures that the interfaces and states of multiple views are consistent. Attention management uses perceptual techniques to focus the user's attention on the right view at the right time. The guidelines aim to help designers avoid unnecessary complexity and minimize the costs of using multiple view systems. They also address trade-offs between different design considerations and provide a framework for evaluating and improving multiple view systems. The authors conclude that these guidelines will be useful for heuristic walkthroughs of both designs and fully implemented systems.