An experiment was conducted to investigate the compartmental mean residence time, (CMRT) of feed residues in segments of gastrointestinal digesta of mature Holstein steers. The objective was to evaluate assumptions that feed residues flow through ruminal digesta as sequential mixing pools having age-dependent (GN) and age-independent (G1) distributed residence times respectively (GN → G1 flow). The basal diet was a semi-tropical hay containing 98 g crude protein and 503 g apparently digestible DM per kg DM. The hay was consumed and feed residues of different size and/or previous digestion from the hay were inserted into the reticulo-rumen (rumen) and abomasum. Marker profiles appearing at the duodenum and faeces were fitted to various compartment models to estimate CMRT. Post-abomasal CMRT did not differ among solutes or feed residues of different size and previous digestion and constituted only 5·8 % of the CMRT for the entire gastrointestinal tract. Markers initially applied to orally or ruminally dosed feed residues exhibited profiles in duodenal digesta and faeces conforming to GN → G1 flow. Previously undigested, masticated feed residues inserted into the dorsal rumen digesta had longer ruminal CMRT in the GN pool but not the G1 pool than did similarly inserted faecal small particles or normally ingested hay. These results support model assumptions of GN → G1 flow within rumen digesta. The results support mechanisms proposed for the GN pool as the ‘lag-rumination pool’ and the G1 pool as the ‘mass action turnover pool’. If further validated, rumen CMRT in cattle could be estimated from marker profiles in more easily obtained faeces to estimate ruminal CMRT required for feed evaluation systems.