Several groupware systems support casual real time interaction over distance by providing periodically updated snapshots of other people's offices. People then monitor these snapshots to determine how available others are for communication. In this research, we try to isolate what information people use from these snapshots to help them infer another's availability. Research participants examined video snapshots of people posed in typical office situations, and judged how available those people were for interaction. Our first result suggests that people have difficulty extracting information from these images unless image resolution was at least 128x128 pixels. Our second result indicates that people interpret stereotypic situations as indicating varying degrees of availability. In general, people are judged as less available when they are seen to be absent from their office, or in conversation with others. People are judged more available when they are in transition (e.g., ...