A rank- and sign-dependent utility theory is based on an operation of joint receipt of 2 independent gambles and 3 assumptions regarding the operation (R. D. Luce, 1992). The authors tested these 3 assumptions-segregation, duplex decomposition, and the additivity of certainty equivalents over joint receipt-using both judged and choice certainty equivalents (CEs). Median choice CEs provided support for both segregation and duplex decomposition but little support for additivity in gains or in losses, whereas median judged CEs also failed to support segregation. The latter failure appears to result from some subjects misunderstanding the instructions.