Abstract.We conducted a usability study on a 50" plasma display comparing three different browser alternatives (iconic, zoom-and-pan, and bifocal view) Large displays are becoming a reality for an increasing number of users as prices go down. So far, browsing techniques have mainly been studied on regular desktop displays.There are many usability issues to consider when display size differs radically from standard size of desktop displays. Freehand interaction with other input devices than the regular 2D desktop mouse, for instance, is a highly relevant part of interaction with large displays.This study is mainly a comparison of browsing techniques, but we also discuss observations made regarding interaction with a large display and a 3D input device. The browser alternatives included in this study are: an iconic browser, a zoom-and-pan browser, and a fisheye browser [7]. Our iconic browser operates in a manner similar to the standard windowing system. Clicking a thumbnail replaces the bulletin board with a readable-sized version of the note. Our zoomand-pan browser was implemented by using the Piccolo toolkit [3] with modified zoom function. Finally, we implemented a type of fisheye browser called the "bifocal view" [13].In order to determine which had the best performance, we conducted a usability study comparing the iconic, zoom-and-pan, and bifocalview browsers on a 50" plasma display with a 3D mouse. Ten subjects completed our study. Each participant used each browser during the experiment. After a review of related work, we will describe the browsers and the study in detail. This will be followed by a discussion of the results and a summary with a discussion of future work.In their image-browser taxonomy, Plaisant, Carr, and Shneiderman [16] describe different types of tasks and how these benefit from different types of browser characteristics.Several browser studies have been conducted, with various results. No single browser has been better overall than the others for all types of tasks.In a study comparing scroll bars, dragging, a pop-up overview, and a pop-up overview containing a field-of-view indicator, Kaptelinin found that the interfaces with a pop-up overview were significantly faster than both the scroll-bar interface and the drag interface [12].Beard and Walker [2] compared interfaces using scrollbars, zoom, and roam (pan) with and without overview, and found that all techniques gave better results with the overview, and scroll bars were significantly slower than the other two techniques.A study by Hornbaek, Plaisant, and Bederson [11] comparing zooming interfaces with and without an overview found that subjects were faster without the overview, but preferred an interface that had one.