Image color saliency has been popularly exploited to detect airports from remote sensing images (RSIs). However, in a complex environment, many non-airport image regions could also produce high saliency, leading to poor robustness of the airport detection result. To address this fundamental problem, a novel airport detection method based on an environment-insensitive saliency analysis process is proposed in the present paper. In our saliency analysis process, an iterative scheme is developed to compute the color saliency of the input RSI, by which environmental disturbances (e.g. mountains, oceans) can be progressively suppressed from the saliency computation though iterative refinement. Since the saliency of non-airport areas is reduced, the target airport can be identified with high robustness. By further fusing the color saliency with the RSI structural saliency, airport support region can be obtained. To capture the airport with high completeness, a region growing algorithm is also proposed, which first selects a set of seed pixels from the saliency map and then greedily adds neighboring pixels with about the same gray-scale value to form a complete profile of the airport region as the final detection output. Extensive experimental results show that our proposed airport detection method can outperform the current state-of-the-art methods by a large margin.