This paper addresses the challenge of infrared small target detection (IRSTD) by proposing a novel Scale and Location Sensitive (SLS) loss function and a simple Multi-Scale Head (MSHNet) detector. The SLS loss function is designed to improve the detection performance by addressing the limitations of existing loss functions, such as the Intersection over Union (IoU) and Dice losses, which lack sensitivity to the scales and locations of targets. The SLS loss includes a scale-sensitive loss and a location-sensitive loss, where the scale-sensitive loss adjusts the weight of the IoU loss based on the predicted and ground-truth scales, and the location-sensitive loss introduces a penalty term based on the center points of targets to enhance localization accuracy. The MSHNet detector is built on a plain U-Net architecture, incorporating a multi-scale head to produce multi-scale predictions for each input. By applying the SLS loss to different scales of predictions, MSHNet achieves superior performance compared to existing state-of-the-art methods, demonstrating a better balance between detection performance, floating-point operations (FLOPs), and inference time consumption. The effectiveness and generalization of the SLS loss are further validated by training existing detectors with this loss function, showing improved detection performance. The code for the proposed method is available at <https://github.com/ying-fu/MSHNet>.This paper addresses the challenge of infrared small target detection (IRSTD) by proposing a novel Scale and Location Sensitive (SLS) loss function and a simple Multi-Scale Head (MSHNet) detector. The SLS loss function is designed to improve the detection performance by addressing the limitations of existing loss functions, such as the Intersection over Union (IoU) and Dice losses, which lack sensitivity to the scales and locations of targets. The SLS loss includes a scale-sensitive loss and a location-sensitive loss, where the scale-sensitive loss adjusts the weight of the IoU loss based on the predicted and ground-truth scales, and the location-sensitive loss introduces a penalty term based on the center points of targets to enhance localization accuracy. The MSHNet detector is built on a plain U-Net architecture, incorporating a multi-scale head to produce multi-scale predictions for each input. By applying the SLS loss to different scales of predictions, MSHNet achieves superior performance compared to existing state-of-the-art methods, demonstrating a better balance between detection performance, floating-point operations (FLOPs), and inference time consumption. The effectiveness and generalization of the SLS loss are further validated by training existing detectors with this loss function, showing improved detection performance. The code for the proposed method is available at <https://github.com/ying-fu/MSHNet>.