Semiempirical AM1 calculations have been carried out on host-guest complexes of model hemicarcerands 1a and 2a. The justification for the choice of the AM1 Hamiltonian was based on a comparison between reported X-ray data for the smaller tetrabromocavitand 4a and computational results obtained using several different Hamiltonians. The complexation behavior of hemicarcerands 1a and 2a have been compared with experimental results reported by Cram et al. for the related hemicarcerands 1b and 2b. Based on this comparison, a criterion for predicting guest encapsulation was developed, E(complexation), which relies on the calculation of AM1 heats of formation for host, guest, and hemicarceplex. If E(complexation) is lower than 10 kcal/mol, then a guest will be encapsulated, while if it is greater than 30 kcal/mol, a guest will not be encapsulated. The use of constrained-path AM1 optimizations to determine the energy barriers to guest entry and exit from the host was found to be a useful tool for examining suitable host-guest combinations when the E(complexation) criteria does not hold. We have computed the barriers to exit of N, N-dimethylformamide (dmf) and furan from the hemicarcerand 1a, the former has been compared with the experiment and shows excellent agreement. Based on the success of the above computational methods in predicting which host-guest combinations will form stable hemicarceplexes we have synthesized a new target hemicarceplex 1b.furan.