Most existing human pose estimation (HPE) methods exploit multi-scale information by fusing feature maps of four different spatial sizes, i.e. 1/4, 1/8, 1/16, and 1/32 of the input image. There are two drawbacks of this strategy: 1) feature maps of different spatial sizes may be not well spatially aligned, which potentially hurts the accuracy of keypoint location; 2) these scales are fixed and inflexible, which may restrict the generalization ability over various human sizes. To- wards these issues, we propose an adaptive dilated convolution (ADC). It can generate and fuse multi-scale features of the same spatial sizes by setting different dilation rates for different channels. Specifically, it uses a regression module to adaptively generate dilation rates for different channels. This also enables ADC to adjust the fused scales according to the sizes of test persons, and thus helps ADC to have better generalization ability. ADC can be end-to-end trained and easily plugged into existing methods. Extensive experiments show that ADC can bring consistent improvements to various HPE methods. The source codes will be released for further research.