AbstractThe detergent system of analysis partitions lignin into two fractions: acid detergent lignin (ADL), usually equated with forage lignin, and ‘acid detergent dispersible lignin’ (ADDL), the existence of which has scarcely been recognized, but can be higher than ADL in tropical grasses. For nine species ADL and ADDL as percentage of original dry matter were, respectively: Aristides calcyna. 8.9 and 9.8; Astrebla squarrosa, 8.6 and 14.7; Chloris gayana, 3.9 and 6.7; Eragrostis sp, 4.3 and 8.8; Heteropogon eontorius, 3.9 and 10.4; Iseilema membranaceum 7.2 and 8.2; Panicum maximum, 1.8 and 5.7; Themeda triandra, 7.3 and 8.2; Sorghum sp, 3.4 and 13.8. Tropical legumes (four species) had much lower levels of ADDL. Azo‐stained fibre showed that most of the free phenolic functions and the alkali‐soluble lignin were in the ADDL fraction. The difference between grasses and legumes was so marked that azo‐staining may provide a method for distinguishing grass and legume particles in feed or faecal samples. The apparent discrepancies between grasses and legumes in the effect of lignin on digestibility may be explained by the grass lignin having been grossly underestimated as ADL. When Heteropogon contortus and Astrebla squarrosa were incubated in dacron bags in the rumen the ratio of ADDL to ADL in the residual fibre decreased markedly in the first 48 h, indicating a higher apparent digestibility for ADDL. From feeding experiments with sheep the apparent digestibility of ADL and ADDL were 18 and 38% for Astrebia squarrosa and 14 and 26% for Stylosanthes hamata. As isolated lignin corresponding to ADDL was inhibitory in vitro to a mixed rumen microbial population at a concentration of 0.12 mg ml−1. Measurement of ADDL could easily be included in fibre analysis by the sequential detergent method as it would require only UV absorbance measurement on the acid detergent filtrate. This would allow measurement of total lignin within the existing system.