Classifications of airborne mutlispectral scanner data for forest defoliation assessment have generally met with only moderate success. Key factors affecting defoliation assessment (radiometric distortions within the imagery due to atmosphere, sun-object-viewer geometry and topography, small spectral differences between defoliation levels, spectral variability due to stand characteristics, and confusion of defoliation classes with other forest types) are examined with the aid of an example of current spruce budworm defoliation in a mixed fir and spruce forest which also contains cumulative spruce budworm defoliation. Analysis indicates that these factors can seriously affect classification of defoliation levels. For example, for certain spectral bands, radiometric distortion can be an order of magnitude larger than the difference between current spruce budworm defoliation levels, slopes of 5° to 15° can cause differences in multispectral scanner intensities equivalent to the range of intensities for areas of healthy to severe current defoliation, and 5 to 10 percent differences in the hardwood component in mixedwood stands can result in intensity differences equivalent to the differences separating defoliation levels. Confusion of current defoliation classes with mixedwood stands and stands of cumulative defoliation can be particularly severe.