The multi-fractal behaviors of relative humidity over China are studied using the multi-fractal detrended fluctuation analysis (DFA) method. Three multi- fractal parameters (the spectrum width, the asymmetry as, and the long-range correlation exponent 0) of the singularity spectrum are introduced to quantify the multi-fractal behaviors. The results show that multi-frac¬tality exists in daily humidity records over most stations in China and is mainly due to the broad distribution of the probability density of the sequence values. Strong multi- fractal behaviors over some stations in the Yunnan, Guang¬dong, and Inner Mongolia provinces are obvious. These behaviors are mainly caused by different long- range correlations between large and small fluctuations. The asymmetry of the singularity of relative humidity records is weak, except for a small number of stations in the far east and west of China, where the singularity spec-trum is left-skewed. Finally, the long-range correlations in North China are stronger than those in South China, which indicates better predictability in North China. By studying the parameters of the multi-fractal spectrum, various data of long-range power law correlations of the relative humidity records are obtained, which may pro-vide theoretical support for climate prediction.