Using radiosonde observations at five stations in the tropical western Pacific and reanalysis data for the 15 years from 2005 to 2019, we report an extremely negative anomaly in atmospheric water vapor during the super El Niño winter of 2015/16 and compare the anomaly with that in the other three El Niño winters of the period. A strong specific humidity anomaly is concentrated below 8 km of the troposphere with a peak at 2.5-3.5 km, and a column-integrated water vapor mass anomaly over the five radiosonde sites has a large negative correlation coefficient of -0.63 with the oceanic Niño3.4 index but with a lag of about 2-3 months. In general, the tropical circulation anomaly in the El Niño winter is characterized by divergence (convergence) in the lower troposphere over the tropical western (eastern) Pacific; thus, the water vapor decreases over the tropical western Pacific as upward motion is suppressed. The variability of the Hadley circulation is quite small and has little influence on the observed water vapor anomaly. The anomaly of the Walker circulation makes a considerable contribution to the total anomaly in all four El Niño winters, especially in the 2006/07 and 2015/16 eastern Pacific (EP) El Niño events. The monsoon circulation shows a remarkable change from one event to another, and its anomaly is large in the 2009/10 and 2018/19 central Pacific (CP) El Niño winters and small in the two EP El Niño winters. The observed water vapor anomaly is caused mainly by the Walker circulation anomaly in the super EP event of 2015/16 but is caused by the monsoon circulation anomaly in the strong CP event of 2009/10. The roles of the Hadley, Walker, and monsoon circulations in the EP and CP events are confirmed by the composite EP and CP El Niños based on the reanalysis data for 41 years. Owing to the anomalous decrease in upward transport of water vapor during the El Niño winter, lower cloud amounts and more outgoing longwave radiation over the five stations are clearly presented in satellite observation. In addition, a detailed comparison of water vapor in the reanalysis, radiosonde, and satellite data shows a fine confidence level for the datasets; nevertheless, the reanalysis seems to slightly underestimate the water vapor over the five stations in the 2009/10 winter.