Introduction In Rajasthan (India) people mainly pay for their healthcare out-of-pocket as they are not insured. The rural poor might pay for health insurance if benefits match their preferences. To find out those preferences in a structured way, we carried out a simulation exercise with a tool called CHAT (Choosing Healthplans All Together). 70 groups participated (1109 individuals), in Churu and Hanumangarh districts of Rajasthan in August-October 2006. This presentation is based on the initial analysis of group choices recorded. Method CHAT is designed as a board-game in which players use stickers to choose healthcare benefits within a pre-defined finite "sticker budget". The actuarial cost of each choice is reflected in the number of stickers needed to "buy" the benefit. Participants can choose between 10 benefit types (hospitalization, consultation, drugs, tests & imaging, indirect costs, preventive care, maternity & family planning, medical equipment, dental care and mental care) at 2 coverage levels (basic - 50% of costs covered; high - 50% of costs covered till a predefined threshold, and 100% above that) . Participants were asked to compose the package they thought was best for their group. Participants were also asked to rank by importance the following three criteria: 'reimbursement' (reimbursement regardless of the absolute level of expenditure), 'fairness' (higher reimbursement rate for higher expenses) and 'catastrophic coverage' (insurance for catastrophic exposure). Results Participants showed a clear preference for catastrophic protection (66%) compared to the reimbursement criterion (20%) and the fairness criterion (14%). The benefits tests & imaging, drugs, hospitalization, indirect costs and maternity are chosen by around 90% of the groups (or more); the first 2 benefits are even chosen by all groups. Out of the four most costly ("major") benefits (hospitalization, consultation, drugs, tests) 65 groups chose hospitalization, drugs and tests, in 5 different combinations when taking into account the coverage levels. These choices were analyzed using data on health care utilization and costs collected in 5 poor locations in India (but not in Rajasthan, for lack of data there). Results show that coverage of drugs at high level offers the best catastrophic protection (for highest decile of costs for a single illness episode), except for the most extreme cases (within the top one percent, where hospitalization at high level offers the best catastrophic protection). However the choices recorded show that despite having stated a preference for catastrophic cover, most groups chose to trade off the high coverage for drugs or hospitalization for coverage of maternity and indirect cost of hospitalization. Conclusions Participants clearly prefer broad coverage over a narrow package with higher coverage. Results strongly suggest that within the finite budget, respondents prioritized a broader package with five benefit types and were willing to forego consultation or high hospitalization coverage. This contrasts with the common notion of insurers offering mainly hospitalization at high level. Although the participants stated a preference for catastrophic protection, most groups included maternity (88%) or indirect costs (94%) rather than upgrading drugs to high coverage level (which gives better results in terms of catastrophic protection)