We report the results of a series of laboratory experiments that focus on a two-person, sequential trust game. Two persons at different locations are paired for a single play of the game. Participants also complete two risk-assessment instruments: a survey and a decision task. The treatments involve the manipulation of the information available to the subjects about their counterparts. In one treatment, subjects are told the sex of their counterpart along with answers to several preference questions (what is your favorite color?, etc.) In the other, each subject sees a photograph of his/her counterpart. Our two-site design means that we can control the information subjects have about their counterparts without compromising anonymity or deceiving the participants. We conjecture that people are strategic actors whose trusting behavior is conditional on the decision context, including the characteristics of a partner, and we are able to test the effect of observable characteristics on trust and reciprocity. In a second component of the experiment, the photographs are rated by a separate set of subjects using survey items that ask subjects to choose between descriptive word pairs. These items scale into three factors that measure trustworthiness/reliability; friendliness/cooperativeness, and mood. Trusting behavior is significantly related to risk attitudes, friendliness/cooperativeness, race and sex. Reciprocity is related to the trustworthiness of the recipient.