Conventional molecular dynamics (MD) simulations are seriously limited by the slow rate of diffusive mixing in their ability to predict lateral distributions of different lipid types within mixed-lipid bilayers using atomistic models. A method to overcome this limitation, using configuration-bias Monte Carlo (MC) "mutation" moves to transform lipids from one type to another in dynamic equilibrium, is demonstrated in binary fluid-phase mixtures of lipids whose tails differ in length by four carbons. The hybrid MC-MD method operates within a semigrand canonical ensemble, so that an equilibrium composition of the mixture is determined by a constant difference in chemical potential (Delta mu) chosen for the components. Within several nanoseconds, bilayer structures initiated as pure dipalmitoyl phosphatidylcholine (DPPC) or pure dilauroyl phosphatidylcholine (DLPC) converge to a common composition and structure in independent simulations conducted at the same Delta mu. Trends in bilayer thickness, area per lipid, density distributions across the bilayer, and order parameters have been investigated at three mixture compositions and compared with results from the pure bilayers at 323 K. The mixtures showed a moderate increase in DPPC acyl tail sites crossing the bilayer midplane relative to pure DPPC. Correlations between lateral positions of the two lipid types within or across the bilayer were found to be weak or absent. While the lateral distribution is consistent with nearly ideal mixing, the dependence of composition on Delta mu indicates a positive excess free energy of mixing.