Experimental data, obtained with an oxygen optical sensor constituted by a polysulfone layer embedding ruthenium(II)(4,7-diphenyl-l,l0-phenanthroline)octylsulfate (Ru(dpp)OS), were rationalized by using the digital simulation technique and generalized for different sensors. The experimental, asymmetric, emission shape was used to define two sensible parameters, ASY (asymmetry factor) and DeltaI(%) (percent variation of emission intensity), to characterize the sensitivity of a generic oxygen optical sensor (represented by the Stern-Volmer constant, K'(sv)). Correlations between ASY and K'(sv) and between DeltaI(%) and K'(sv) were established, and a double working curve was proposed to evaluate with a single light emission measurement the K'(sv) value with the best precision. Sensitive membranes (-log K'(sv) = pK'(sv) < 0.5) had high precision only for low %O(2) values; poorly sensitive membrane (pK'(sv)> 2.5) had constant but scarce precisions in a large %O(2) interval. For %O(2) up to 21% (air) good values are pK'(sv)= 0.5-1.0. In order to monitor a wider %O(2) range, pK'(sv) = 1.5-2.0 are good choices. A simple mathematical model allowed one to estimate the oxygen diffusion coefficient inside the layer, D(O2), and its solubility in the polymer matrix, s(O2), from the simple measurement of the membrane thickness, response time, t(90), and luminescence lifetime. D(O2) = 2 x 10(-8) cm2 s(-1) and s(O2) = 2.2 x 10(-3) mol atm(-1) dm(-3) [corrected] were estimated for our membranes. The proposed working curves gave very good results even with literature data.