Since large-scale variations in photosynthetically active radiation (PAR) influence the terrestrial carbon sources and sinks through the plant photosynthesis variations, large-scale evaluation of PAR is required. In the present study a simple PAR estimation model was developed for Southeast Asia, where large-scale forest fires occurred during El Ni¿o years. The model considered the smoke aerosol released by forest fires using satellite-based smoke detection methods. A comparison study with ground-based solar radiation data for Malaysia and Thailand indicated that the current model could estimate monthly PAR with 10% (root-mean-square) accuracy and would successfully trace the seasonal and year-to-year variations in PAR, including the forest fire periods. During the peak-smoke month in Indonesia, September 1997, the reduction of PAR by smoke reached 63--75% in the center of the Kalimantan and Sumatra Islands. From the analyses of the smoke and cloud cover impacts on PAR in 1997--1999, annual PAR variations were found to be mostly regulated by smoke variations on the Malay Peninsula, Sumatra, and Kalimantan Islands and cloud cover variations on the Indo-China Peninsula. Thus annual variations in PAR changed with location. These variations did not simply correlate with year-to-year variations in cloud cover associated with the El Ni¿o and La Ni¿a cycle, but exhibited more complicated spatial variations due to the existence of smoke in Southeast Asia.