Reference evapotranspiration (ET_0) is the water requirement of plants driven by meteorological factors. It is critical for restoration of vegetation and regional water resources management. In Southwest China where karst landforms are common, the ecological environment is fragile. Extreme drought and flood events are usually caused by declining annual precipitation and large amounts of precipitation within short time periods. The response of ET_0 to global climate change is a key in the assessment of the effect of climate change on hydrology. Therefore, it was important to explore the spatial and temporal variations and the related driving factors of ET_0. Based on data collected from 25 meteorological stations in Guangxi for the 1960-2010, daily ET_0 was computed using the FAO-56 Penman-Monteith equation. The spatio-temporal variations and the driving factors of ET_0 in Guangxi karst and non-karst areas were analyzed separately using Kriging interpolations, cumulative anomalies and spearman rank correlation at seasonal and annual scales. The driving factors of ET_0 in karst and non-karst areas were analyzed using the path-analysis method. Results showed that ET_0 decreased gradually from southeast to northwest of the study area, with an average rate of 1 138 mm*a~(-1) during the studied period. High ET_0 was mainly distributed in non-karst areas and low ET_0 in karst areas. For different seasons, ET_0 values from stations in non-karst areas were generally higher than those from stations in karst areas. Mean monthly ET_0 in both karst areas and non-karst areas tracked a unimodal curve, with relatively high ET_0 for the period from May to September. Non-karst areas had higher ET_0 variations than karst areas. The average monthly ET_0 in non-karst areas was 95 mm and that in karst areas 64 mm. Summer ET_0 was highest and winter ET_0 lowest. In the recent 51 years, annual ET_0 for both karst and non-karst areas tracked N-shaped distribution, and were the highest in the 1970s and lowest in the 1990s with an increasing trend for period 2002-2010. The trends of variation in annual ET_0 in Guangxi were also analyzed. Out of the 25 stations, 3 stations in karst and 4 stations in non-karst areas had significant decrease in trend; then 2 stations in karst and 1 station in non-karst areas showed significant increase in trend, and the other stations had no significant trends. Analysis of the relationship between ET_0 and meteorological factors showed that sunshine duration, wind speed and average temperature were the main meteorological factors driving ET_0 in non-karst areas. Relative humidity influenced annual ET_0 by interacting with other meteorological factors in karst areas. Seasonally, sunshine duration and average temperature had the highest impact on ET_0, and both were positively related with ET_0. Furthermore, the indirect effect of the coefficient between wind speed and ET_0 was negative for karst areas and positive for non-karst areas in winter and spring. Differences in geographical location and topography were the main reasons for the differences in ET_0 between karst and non-karst areas. Therefore, understanding the trend of change and the reasons for the change in ET_0 in different regions was a necessary measure for analysis of ecological water requirement.