(Go back one page, or back to the Introduction).
For the most part, autofluorescence does not affect compensation. Autofluorescence is present in all channels, and it will still be present after compensation (although the absolute value will be altered by the compensation). For most applications, we can ignore the contributions of autofluorescence to compensation; we do need to subtract it out to compute antigen densities (i.e., autofluorescence is a background). Finally, note that the autofluorescence of the positive and negative populations must be the same in order to achieve proper compensation! In other words, using a fluorescence probe specific to monocytes and lining them up with lymphocytes, whose autofluorescence is significantly lower than monocytes, will fail to yield proper compensation.
In general (for PBMC analyses), the best compensation is attained by mixing, in roughly equal proportions, unstained cells with an aliquot of the same cells stained with the appropriate antibodies--and then, when setting the compensation, to use a lymphocyte gate when viewing the populations. In this case, autofluorescence can be entirely ignored. Using a compensation stain which divides lymphocytes (e.g., CD4, CD8, or CD19) into subsets is acceptable, with the caveat that proper compensation assumes that the positive and negative populations (in this case, B and T cell subsets) have equal levels of autofluorescence. Note that if a lymphocyte gate is not used, then the fluorescence of the positively stained cells (which are only lymphocytes) will be matched against the negatively stained cells (which are lymphocytes and monocytes), with the result that the sample will be undercompensated.