OBJECTIVES: Using (SG) or time tradeoff (TTO) measures of QALY utility weights requires strong and generally unrealistic assumptions that effectively break the link between SG and TTO utilities and utility-theoretic preferences. In particular, QALY linearity assumptions violate the generally accepted law of diminishing marginal utility. This presentation derives general "super QALYs" from nonlinear, ordinal utility functions that incorporate wealth and non-health utility variables as well as time in specific health states. Thus super QALYs can incorporate patient satisfaction related to dosage or drug administration and other utility-relevant factors for both acute and chronic health outcomes. METHODS: We illustrate comparisons between conventional QALY and super QALY measures using linear and Cobb-Douglas utility functions. Under general conditions, nonlinear super QALYs may rank health outcomes differently than QALYs, but linear super QALYs rank outcomes similarly. If preferences are nonlinear, marginal changes will be weighted differently with QALYs and super QALYs, thus affecting incremental cost-effectiveness ratios. When derived from consistent utility-theoretic preference relations, super QALYs, willingness to pay (WTP), and willingness to wait (WTW) represent alternative and equivalent rescaling of the same preference information. RESULTS: We explore the empirical significance of these conceptual results using an empirical utility function from a recent choice-format conjoint study of acute respiratory and cardio-vascular symptoms. Setting utility to zero for the worst-observed outcome, quality-adjusted life days (QALDs) range from 0.066 for 5 days of pneumonia symptoms requiring hospitalization to 4.25 for 5 days of nasal congestion with some physical activity restrictions. Corresponding WTP, WTP per QALD, and WTW values for the same outcomes are $980 and $30, $199 and $40, and 4.93 and 0.75, respectively. CONCLUSIONS: Although nonlinear ordinal utility functions sacrifice the simplicity of constant QALY weights and linearity over time, improved validity justifies the modest increase in analytical burden in many cases.