Pedestrian detection is a critical task in computer vision because of its role in ensuring traffic safety. However, existing methods that rely solely on RGB images suffer from performance degradation under low-light conditions due to the lack of useful information. To address this issue, recent multispectral detection approaches combine thermal images to provide complementary information. Nevertheless, these approaches have limitations such as the noisy fused feature maps and the loss of informative features. In this paper, we propose a novel target-aware fusion strategy for multispectral pedestrian detection, named TFDet. Unlike existing methods, TFDet enhances features by supervising the fusion process with a correlation-maximum loss function. Our fusion strategy highlights the pedestrian-related features while suppressing the unrelated ones. TFDet achieves state-of-the-art performances on both KAIST and LLVIP benchmarks, with a speed comparable to the previous state-of-the-art counterpart. Importantly, TFDet performs remarkably well under low-light conditions, which is a significant advancement for road safety.