Global data sets of surface-incident photosynthetically active radiation (PAR, 400--700 nm) from three satellite-based methodologies are compared with each other and with ground data. Two are based on C1 data from the International Satellite Cloud Climatology Project (ISCCP): ISCCP-BR was produced from an existing total shortwave irradiance data set by assuming 50% PAR; ISCCP-PL is an existing PAR-specific data set. TOMS PAR is based on 370 nm data from the Nimbus-7 Total Ozone Mapping Spectrometer (TOMS). ISCCP-BR and ISCCP-PL are typically 12--16% and 8--12% greater than TOMS PAR, respectively. Comparison of ISCCP-BR, ISCCP-PL, and TOMS PAR with ground data at a mid-latitude site for non-winter (snow/ice-free) months yields RMS differences of 28.1%, 13.7%, and 7.2%, and biases of +25.9%, +12.0%, and +2.8%, respectively. The comparison indicates TOMS PAR provides the most reliable estimates of PAR for snow/ice-free, growing season months, i.e., when the energy is effective for photosynthesis. Differences among the data sets constitute a significant source of variation in results from biospheric process models that include PAR as a variable. ¿ American Geophysical Union 1995