The low-energy conformers of alanine and glycine dipeptide have been modeled in the gas phase and in aqueous solution using ab initio methods. In the gas phase, seven low-energy minima have been located for the alanine dipeptide (AD), compared to six for the analogue (ADA), without terminal methyl groups. For the glycine dipeptide (GD), four minima are found, compared to two for the corresponding analogue (GDA). The effect of solvent has been included using both the self-consistent reaction field (SCRF) and polarized continuum (PCM) methods. Calculations of the solvated dipeptide were performed using both the gas-phase and the SCRF-optimized structures. For alanine dipeptide, solvation calculations performed with the free molecule-optimized structures using the PCM predicts the C5 conformation to be the most stable, which is not in agreement with the limited experimental data or with molecular dynamics simulations. For the SCRF model, using free molecule-optimized structures, the beta(2) conformation is predicted to be the most stable with the C5 the next most stable. Optimization of the alanine dipeptide conformations within the SCRF model still predicts the beta(2) conformation to be the most stable. However, the beta conformation is only slightly higher in energy, while the alpha(R) conformation is not a stationary point. Application of the PCM method to the SCRF-optimized structures reverses the conformational preference of beta(2) and beta. For glycine dipeptide, using free molecule-optimized structures, the PCM method predicts C5 to be the most stable conformation, while for the SCRF method the C5 and beta(2) conformations are predicted to be the most stable. Optimization of the glycine conformations with the SCRF method results in only two conformations, the modified left- and right-handed cr conformations which are equivalent in energy. A comparison of these results with those from explicit inclusion of solvent molecules is made.