The multispectral image analysis technique to detect chlorophyll fluorescence (CF) in soybean seeds was studied to assess the relationship between CF signals and seed physiological potential. Eight treatments, corresponding to 0%, 2%, 4%, 6%, 8%, 10%, 12%, and 14% green seeds, were used on two cultivars, BMX Desafio RR 8473 RSF and 96R10 IPRO, which passed through different seed quality tests. Initially, the CF of the seeds was determined using 660 nm and 730 nm spectra, and then the germination, electrical conductivity, accelerated aging with saturated NaCl solution, tetrazolium, and computerized seedling image analysis (Vigor-S) tests were performed on the same seeds. A completely randomized design was used, as well as replications of each treatment. Analysis of variance (ANOVA) was performed on the data from germination, vigor, and CF tests using the R® software, and the means were grouped by the Scott-Knott test (p ≤ 0.05). Pearson’s linear correlation coefficients (r) were calculated for all combinations among the evaluations with significance of the r values determined by the t-test (p ≤ 0.05), and multivariate analysis of the principal components was performed. Proportional increases in green seeds contribute to an increase in chlorophyll fluorescence signals and have a negative correlation with seed physiological quality; levels above 4% green seeds in the samples result in marked losses in physiological potential. Therefore, the chlorophyll fluorescence detected through multispectral images is inversely related to the physiological potential of soybean seeds.