Fibrobacter succinogenes S85 digested ballmilled cellulose at a rapid rate (0.10 h(-1)), but there was a long lag time if the culture was not transferred daily. When F. succinogenes was starved for 100h, a large fraction of the cells (>30%) still bound to cellulose, but the lag time was 150 h. The lag time was similar for either cellulose- or cellobiose-grown inocula, and lag times were highly correlated (r(2) = 0.91) with a decrease in viable cell number. The number of viable cells declined from 10(8) to 10(6) in the first 30h of starvation, and by 72h the viable cell number was less than 10(3)/ml. Cells growing exponentially on cellobiose had a large pool of polysaccharide, and continuous culture experiments indicated that polysaccharide accumulation was not significantly influenced by the growth rate of the culture (approximately 0.7 mg polysaccharide mg(-1) protein). When the cellobiose was depleted, cellular polysaccharide decreased at a first order rate of 0.09 h(-1) The rate of endogenous metabolism was initially 0.08mg polysaccharide mg(-1) protein h(-1), and there was little decline in viability until the rate of endogenous metabolism was less than 0.01mg polysaccharide mg(-1) protein h(-1). When the rate was less than 0.01 mg polysaccharide mg(-1) protein h(-1), the cells could not maintain a sodium gradient, transport cellobiose or grow. The endogenous metabolic rate needed for cell survival was 20 fold less than the maintenance energy of cells growing in continuous culture (0.01 versus 0.232 mg carbohydrate mg(-1) protein h(-1)).