Twenty fine-wool, ruminally cannulated lambs (average weight 45.9 kg) were used in a completely random design to evaluate the ability of three internal markers to predict dry matter digestibility and two external markers to estimate faecal output. Lambs were allotted randomly to one of four diets: 100 % prairie hay (PH), 100 % lucerne hay (LH), 50 % prairie hay: 50 % sorghum grain (PS) and 50% lucerne hay: 50% sorghum grain (LS). The trial consisted of a 14-day adaptation period followed by a 7-day total faecal collection period. Feed and faecal samples were subjected to 96 h ruminal fluid and 48 h acid-pepsin digestions, followed by extraction with acid detergent (IVADF) or neutral detergent (IVNDF) solution. Dry matter digestibility (DMD) calculated from feed:faeces ratios of IVADF, IVNDF and acid detergent lignin (ADL) was compared with in vivo apparent digestibility. Ytterbium-labelled forage (YLF) and dysprosium-labelled faeces (DLF) were pulse-dosed via ruminal cannulae, and faecal Yb and Dy excretion curves were fitted to a one-compartment, age-dependent model for estimation of faecal output, particulate passage rate (PPR) and mean gastrointestinal retention time. In vivo DMD in lambs fed PH was greater (P < 0.05) than DMD calculated from IVNDF, IVADF and ADL. In lambs fed LH and LS, in vivo DMD did not differ (P > 0.05) from marker estimates. In vivo DMD for lambs fed PS did not differ from IVNDF or IVADF estimates but was greater than (P < 0.05) the ADL estimate. No differences (P > 0.05) were observed in recovery among the three internal markers for any of the diets. Faecal output for lambs fed PH did not differ (P > 0.05) from marker estimates but was overestimated by 15 to 20% by YLF and DLF. Faecal output for lambs fed LH was similar to the estimate from YLF, but less than (P < 0.05) the estimate with DLF. For lambs fed PS, faecal output did not differ from marker estimates, but YLF and DLF values were 16% lower and 17% higher, respectively. No significant differences were observed in actual and estimated faecal output for lambs fed the LS diet. Estimates of PPR with DLF were numerically greater than YLF estimates for all diets except LS. Correspondingly, mean gastrointestinal retention time was less (P < 0.05) for DLF compared with YLF for all diets except LS.