In this study, all‐sky ShortWave Direct Aerosol Radiative Forcing (SWDARF) at the top of atmosphere is estimated using the method of Oikawa et al. (2013) applied to two generations of Cloud‐Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) Level 2 products, i.e., version 2 (V2) and version 3 (V3), and the Moderate Resolution Imaging Spectroradiometer (MODIS) cloud product. The estimated SWDARF in Oikawa et al. (2013) was based on CALIPSO V2 product, which contained significant errors in cloud clearing and low‐altitude aerosols. This error was corrected in V3, resulting in greatly improved and significantly different aerosol and cloud distributions. In clear‐sky conditions, the magnitude of aerosol optical thickness underestimation becomes smaller and SWDARF becomes more negative using the V3 product. In addition, above‐cloud aerosols, which cause positive SWDARF, are less frequently detected and below‐cloud aerosols are more frequently detected in the V3 product than in the V2 product, so that cloudy‐sky SWDARF becomes more negative using the V3 product. From these results, clear‐sky, cloudy‐sky, and all‐sky SWDARFs become more negative using the V3 product than the V2 product. The magnitude of negative SWDARF using the V3 product is more than twice as large as the V2 product under all‐sky conditions due to V3 improvements in the lidar retrieval algorithms. Considering the uncertainties of aerosol and cloud measurements, annual zonal averages of clear‐sky, cloudy‐sky, and all‐sky SWDARFs from 60°S to 60°N are estimated as −4.0 ± 0.2, −1.1 ± 0.3, and −2.1 ± 0.2 Wm−2 from the V3 product.