Poly-L-proline (PLP) polymers are useful mimics of biologically relevant proline-rich sequences. Biophysical and computational studies of PLP polymers in aqueous solutions are challenging because of the diversity of length scales and the slow time scales for conformational conversions. We describe an atomistic simulation approach that combines an improved ABSINTH implicit solvation model, with conformational sampling based on standard and novel Metropolis Monte Carlo moves. Refinements to forcefield parameters were guided by published experimental data for proline-rich systems. We assessed the validity of our simulation results through quantitative comparisons to experimental data that were not used in refining the forcefield parameters. Our analysis shows that PLP polymers form heterogeneous ensembles of conformations characterized by semirigid, rod-like segments interrupted by kinks, which result from a combination of internal cis peptide bonds, flexible backbone psi angles, and the coupling between ring puckering and backbone degrees of freedom.