The effects of leakage, detector and source perturbation, and the presence of higher modes in the neutron density distribution on a determination of the mean lifetime of thermal neutrons in water are discussed. The methods used in several recent experiments to minimize these sources of error are analyzed, with particular attention paid to the problem of suppressing the higher modes of the neutron density distribution. The effect of moderator dimensions is presented in terms of mode suppression factors for three characteristic moderator sizes. Finally, the mathematical analysis for a proposed large-geometry, high precision mean lifetime experiment is presented, in which the neutron distribution is calculated as a solution to an eigenvalue problem with variable boundary conditions. Three approximations are presented which allow the counter perturbation to be calculated and the mode content controlled.