Autofluorescence spectra have been obtained on beef longissimus dorsi muscles (n = 151), and the spectra regressed against Warner-Bratzler (WB) peak values. The spectra obtained depended on the method used, and it is suggested that the difference is related to the use of two different types of Xenon lamps (a pulsed versus a continuous light source) and the inherent kinetic differences in the collection of the fluorochromes' emitted light. Poor to good (R=0.45-0.84) correlations between WB peak values and the emission spectra were obtained. This relationship is established using chemical information originating largely from collagens. Minor differences in predictability were observed using either excitation wavelengths 332 or 380 rim. The emission wavelengths containing the most relevant information about WB peak values were between 360 and 500 nm. Wavelengths around 375 nm, excitation 332 nm, were in particular important and were related to a component in the perimysial tissue, most likely being present in collagen I or III. Excitation at 380 nm revealed the wavelength range 460-480 nm as important presumed due to collagens in the perimysium. An experiment, simulating industrial routines, using 45 samples collected at the slaughterhouse two days post mortem, was carried out. However, for those samples no model was observed between the emission spectra and WB peak values. Only when some samples having very low (< 1.45 mum) sarcomere lengths were removed, could a significant model be obtained. It is concluded that with the prevailing early post mortem routines in slaughterhouses with, among other things, wide variation in chilling rates between muscles and carcasses, the technique of autofluorescence has limited industrial potential as a sole, rapid non-destructive method for measuring tenderness of the longissimus dorsi muscle (LD). (C) 2002 Published by Elsevier Science Ltd. All rights reserved.