In view of hyperspectral remote sensing image classification, this paper introduces Limit learning theory and proposes a novel classification approach for a hyperspectral image (HSI) using a hierarchical local receptive field (LRF) based extreme learning machine (ELM). Considering the local correlations of spectral features, hierarchical architectures with two layers can potentially extract abstract representation and invariant features for better classification performance. Simultaneously, the influence of different parameters of the algorithm on classification performance is also analyzed. Experimental results on two widely used real hyperspectral data sets confirm that the comparison with the current some advanced methods, and the proposed HSI classification approach has faster training speed and better classification performance.