Two different approaches to relate wheat yield with spectral indices derived from remotely-sensed data have been explored for the state of Punjab, India. In the study based on site-level approach yield obtained from crop-cutting sites was found to be linearly related to NIR/Red ratio derived from Landsat MSS data of corresponding sites in Ludhiana and Patiala districts of Punjab. Incorporation of agrometeorological data was also tried. Certain inherent limitations of the site-level approach led to the district-level studies which focused on the relation of district yields with corresponding average spectral indices derived from satellite sensors like Landsat MSS and lRS-LISS-i. Significant correlations were observed in all cases and the relation based on Landsat MSS/IRS LISS-I data was used for trial forecast of wheat yields for 1989–90 season. A comparison of remote-sensing based production forecast showed good agreement with the conventional estimate of Bureau of Economics and Statistics at state level although at district level, deviations were larger.