This paper presents state-of-the-art techniques which enhance noticeably the efficiency of evolutionary algorithms in aerodynamic shape optimization and, in particular, in turbomachin-ery blade airfoil design problems. These techniques rely upon the combined use of hierarchical (more than one levels of search, using evaluation tools of different modeling accuracy and CPU cost), distributed (simultaneously evolving population subsets, allowing the regular exchange of promising solutions between them), metamodel-assisted (many non-promising individuals, generated during the evolution, are filtered out during the inexact pre-evaluation phase, using on-line trained artificial neural networks) evolutionary algorithms. Their combination in a single search method, abbreviated to HDMAEA or HD(EA-IPE) and ported on a multiprocessor computing platform, is presented. This method is applied to the design of the stator airfoil of a highly-loaded compressor cascade. A flow turning of 45° at transonic flow conditions is achieved with markedly low total pressure losses.