This report describes a new electronic atlas of human cerebral cortex that provides a substrate for a wide variety of brain-mapping analyses. The Population-Average, Landmark- and Surface-based (PALS) atlas approach involves surface-based and volume-based representations of cortical shape, each available as population averages and as individual subject data. The specific PALS-B12 atlas introduced here is derived from structural MRI volumes of 12 normal young adults. Accurate cortical surface reconstructions were generated for each hemisphere, and the surfaces were inflated, flattened, and mapped to standard spherical configurations using SureFit and Caret software. A target atlas sphere was generated by averaging selected landmark contours from each of the 24 contributing hemispheres. Each individual hemisphere was deformed to this target using landmark-constrained surface registration. The utility of the resultant PALS-B12 atlas was demonstrated using a variety of analyses. (i) Probabilistic maps of sulcal identity were generated using both surface-based registration (SBR) and conventional volume-based registration (VBR). The SBR approach achieved markedly better consistency of sulcal alignment than did VBR. (ii) A method is introduced for 'multi-fiducial mapping' of volume-averaged group data (e.g., fMRI data, probabilistic architectonic maps) onto each individual hemisphere in the atlas, followed by spatial averaging across the individual maps. This yielded a population-average surface representation that circumvents the biases inherent in choosing any single hemisphere as a target. (iii) Surface-based and volume-based morphometry applied to maps of sulcal depth and sulcal identity demonstrated prominent left-right asymmetries in and near the superior temporal sulcus and Sylvian fissure. Moreover, shape variability in the temporal lobe is significantly greater in the left than the right hemisphere. The PALS-B12 atlas has been registered to other surface-based atlases to facilitate interchange of data and comparison across atlases. All data sets in the PALS-B12 atlas are accessible via the SumsDB database for online and offline visualization and analysis.