The fluorodinitrobenzene (FONB), succinic anhydride (SA), dansyl chloride (DAN), dye-binding lysine (OBL), total lysine (TL), ninhydrin (NIN) and Tetrahymena lysine (TET) methods were compared for their ability to assess available lysine in soyaprotein heated in the absence or presence of glucose, lactose or xylose and in formaldehyde-treated lactalbumin. The reactive lysine methods showed comparable sensitivity to lysine damage in soyaprotein heated in the absence of sugar, the results indicating the presence of acid labile isopeptides and unidentified acid stable derivatives. Results for soyaprotein heated with glucose, lactose or xylose showed that the type of sugar and the extent of heat treatment has a strong influence on the progress of the Maillard reaction. Furthermore since fructoselysine (F-L) and lactulosyl-lysine (L-L) are colourless up to 30% loss of available lysine can occur without any change in product colour. The FONB method is the most sensitive for mildly damaged glucose-soya samples followed by DAN or OBL, SA and TL whereas for mildly damaged lactose-soya samples the order is OBL, FONB, SA, TL and DAN. For severely damaged samples the DAN or SA methods were the most sensitive followed by OBL, FONB and TL. Formylation of lactalbumin occurred more readily at higher formaldehyde concentrations. Exposure time had less effect while pH (5 and 9) had no effect. Methylene derivatives reached maximum levels sooner than the methylol compounds. Lysine and tyrosine but not histidine formed methylene bridges while tyrosine was found to condense with free formaldehyde during acid hydrolysis raising questions as to the interpretation of similar studies reported in the literature. The FONB, OBL and DAN methods were all very sensitive to this type of damage with the NIN and TL methods being less sensitive and the SA method being completely unsuitable. The TET assay is unsuitable for 'early' Maillard damage since at low sample-N levels growth is stimulated by its ability to utilise unavailable F-L and L-L while at higher N-levels growth is inhibited. No single method is most suitable for all types of damage. Furthermore, all except DAN and DBL are either too long, rather complicated, require expensive equipment or involve the use of dangerous chemicals. The DAN method appears promising but the problem of converting arbitrary fluorescence units to lysine values needs to be overcome. The DBL is recommended for routine analysis since it is simple, economical and highly sensitive to all lysine damage provided care is taken to optimise dye-binding for each type of material analysed. / Thesis (Ph.D.)-University of Natal, Pietermaritzburg, 1985.