The polarized‐fluorescence experiment for measurement of preferred orientation is analyzed mathematically. The intensity is dependent uponPandA, unit vectors describing the positions of the polarizer and analyzer, and upon the orientation distributionN(&agr;, &dgr;) of the chromophoric group of the polymer. In the case of biaxial or uniaxial symmetry, the intensityIPAfor any combination ofPandAvectors is expressed as linear combination of nine fundamental intensities characteristic of a given sample. These intensities form a 3×3 matrix, each entryIijbeing the intensity obtained withPandAparallel to theith andjth principal axes of the sample. Of the nine components ofI, only five are independent for a biaxially oriented sample; two for uniaxial orientation. EachIijis proportional to ⟨xi2xj2⟩Av, wherexiandxjare the components of the chromophore vector along the corresponding principal axes. Calculated matrices are shown for several model distributions. Nishijima'sPfunction is related to theIij's, and the problem of measuring biaxial orientation is discussed. TheIij's are also shown to be directly related to the second‐and fourth‐order coefficients in the spherical harmonic expansion of N(&agr;, &dgr;). These coefficients can be measured directly by a few simple experiments involving spinning the sample between crossed or parallel polarizers. Thus, the experimental error in their determination can be minimized.