Although equity is an important consideration in resource allocation decisions, its definition remains contested. We used conjoint analysis to determine the attributes of equity and their relative importance. We recruited patients and their companions from waiting rooms of a large urban teaching hospital. We designed a computer-administered dichotomous choice conjoint analysis survey, consisting of 7 attributes, each with 2 to 3 levels. We generated scenarios using a fractional factorial design consisting of 9 choices of 2 programs each, presented in random order. We asked respondents to consider that they were voting in hypothetical referenda for new health care programs, characterized by the average patient who would benefit. We started with an additional learning choice, which was not analyzed, and finished with an additional consistency choice, in which respondents decided between the best and worst possible combination of attribute levels. We analyzed results using random-effects probit generalized estimating equations, to account for repeated measures within subjects. Thus, we quantified equity as a function of potential attributes. For interpretability, we rescaled this equity function from 0 to 100 representing our worst and best combination of attribute levels. The average age of respondents (n=287) was 45 (95% confidence interval[95%CI] 43,46), 47% were women, 80% had at least some college education, 18% rated their health as bad or fair, 26% made under $40,000/year, 68% were Caucasian, and 59% were working full-time. Each interview lasted 13 (95%CI 13.0,14.3) minutes. We dropped responses from 2 participants who made inconsistent choices. Each 1 point decrease (scale 0 to 100) in baseline quality of life increased the equity function by 0.61 (95%CI 0.56, 0.67). We calculated corresponding values for: baseline life expectancy 0.81 (0.74,0.88) per year decrease; increased life expectancy 3.14 (2.89,3.44) per year; pre-program financial endowment -0.31 (-0.29,-0.34) per $1000; and the age of program recipients -0.26 (-0.24,-0.29) per year. The expected gain in quality of life (p=0.07) and the number of individuals expected to benefit (p=0.36) did not significantly influence the equity function. No demographic characteristics of respondents were significant. Members of the general public preferred to allocate resources to groups with poor baseline life expectancy and quality of life, the young, those with the greatest potential for increased survival, and those who have had less access to health care resources.