An integrated visual thesaurus and results browser to support information retrieval was designed using a task model of information searching. The system provided a hierarchical thesaurus with a results cluster display representing similarity between retrieved documents and relevance ranking using a bullseye metaphor. Latent semantic indexing (LSI) was used as the retrieval engine and to calculate the similarity between documents. The design was tested with two information retrieval tasks. User behaviour, performance and attitude were recorded as well as usability problems. The system had few usability problems and users liked the visualizations, but recall performance was poor. The reasons for poor/good performance were investigated by examining user behaviour and search strategies. Better searchers used the visualizations more e!ectively and spent longer on the task, whereas poorer performances were attributable to poor motivation, di$culty in assessing article relevance and poor use of system visualizations. The bullseye browser display appeared to encourage limited evaluation of article relevance on titles, leading to poor performance. The bullseye display metaphor for article relevance was understood by users; however, they were confused by the concept of similarity searching expressed as visual clusters. The conclusions from the study are that while visual user interfaces for information searching might seem to be usable, they may not actually improve performance. Training and advisor facilities for e!ective search strategies need to be incorporated to enhance the e!ectiveness of visual user interfaces for information retrieval.