This paper proposes a principal-agent framework to study the optimal transfer of longevity risk between a reinsurer and a hedger under information asymmetry.  Most hedgers, such as a defined-benefit pension plan or a life insurer, in the real world have rather small portfolios, the liabilities of which are hard to be accurately estimated by the reinsurer. Using indemnity longevity swaps as an example of reinsurance products, we derive the analytical solution to the optimal risk premiums and incentive-compatible hedge demands in a separating equilibrium and examine the conditions for the existence of the separating equilibrium. The theoretical results are evaluated using real-world mortality data in extensive empirical analyses. We find that the expected profit of the reinsurer can be substantially increased if the adverse selection issue in longevity risk transfer is appropriately addressed.