Abstract: The paper aims to improve the forecast level of visibility and reduce the harm caused by low visibility. The authors used the mathematical statistics and correlation coefficient method to analyze the variation characteristics of visibility and the correlation between the variation of visibility and the meteorological factors by using the horizontal visibility and surface meteorological data from 1991 to 2015 of Weihui. The results showed that: the inter-annual variation trend first decreased and then increased in the 25 years, the low-value area appeared in the early 21st century; the seasonal variation value was best in spring, worst in winter; the lowest value appeared in December and the highest in May; the daily variation value was poor in the morning and after 17:00, and was best between 13:00 to 15:00; the frequency of atmospheric visibility and low atmospheric visibility had an upward trend before 2000, and a fluctuated downward trend after 2009; and the highest frequency appeared in December, the lowest frequency appeared in July and May; the visibility and relative humidity had significantly negative correlation, on the other hand, it had significantly positive correlation with wind speed, and it had no significant correlation with temperature and air pressure.