Two recent court decisions disagreed on question of sex in pension plans, one partially based on a factual error derived argument. It is shown mathematically that 'actuarial fairness' is only weakly related to degree of mortality distribution overlap, and it is pointed out that 'fairness' with respect to a large group of insured may not lead to stable conditions, if ex ante identifiable subgroups differ in mortality. Institutional or moral barriers are then needed to uphold uniform treatment. A parallel is drawn to life annuities for substandard lives. The introduction of unisex mortality tables in pension plans remains an open question. Two circuits of U.S. Court of Appeals rendered confliting decisions. On Sep. 29, 1982 Second Circuit enjoined Teachers Insurance and Annuity Association (TIAA) from using sex-differentiated mortality tables when calculating amount of monthly benefits to which its participants are entitled upon retirement. [1]. On Oct. 14, 1982 Sixth Circuit held that the use of sex-segregated mortality tables to calculate retirement annuities does not violate Title if actuarial value of annuities is equal for similarly situated women and men. [2]. This unstable situation did not last for long; on July 6th, 1983 Supreme Court decided with a 5 to 4 majority that in employment-related pension plans discrimination on basis of sex [is] in violation of Title VII [3]. The matter is still not settled, however; remarks in Court opinions point towards Congress, where a measure (S 372) to eliminate sex in all insurance has been introduced. These problems have prompted discussions that strike to core of concept of insurance, and arguments deserve close scrutiny. A major reason for decision of Second Circuit was overlap argument: sex segregation is unjust because lifetimes for men and women after age 65 follow substantially overlapping distributions. This argument Hanne D. Christiansen is an Assistant Professor of Business at University of Kansas. She graduated University of Copenhagen, taught at Department of Insurance Mathematics, University of Copenhagen, was an Associate Professor of Statistics and Operations Research at Techical University of Denmark, and a Statistician at Danish Hospital Institute. She is a member of Danish Society of Actuaries and of American professional societies. She has published in this Journal and several other professional journals, as well as in two leading Danish newspapers. The paper was supported in part by a grant The University of Kansas General Reseach Fund.