Objectives of this study were 1) to measure small-particle and liquid mean retention time (MRT) of 12 grass hays similar in NDF (61.3 +/- 1.9% NDF) but differing in morphological composition and to relate passage rates to proportions of blade, sheath and stem and 2) to evaluate the influence of MRT of small and large blade and stem fractions and large sheath fractions, in addition to morphological composition, on intake and digestibility of the 12 hays. In each of two periods, 24 sheep (wethers) were offered one of 12 hays at three consecutive levels of feeding: (L1) ad libitum, allowing 15% refusal; (L2) restricted to 100% of hay consumed ad libitum by an individual wether during L1; and (L3) 1.8% of BW on a DM basis. Hays offered included two sorghum-sudan, four barley, four oat and two pearl millet. Marked particles were pulse-dosed in L2 and L3. Large and small particles of stem and leaf were extracted with neutral detergent and marked with Cr or rare earth metals. Marked large and small stem particle MRT generally were longer (P less than .05) than those of corresponding large and small blade particles. Large sheath particles generally had an intermediate (P less than .05) MRT between those of large stem and blade particles. Mean retention times of marked fractions were lower (P less than .05) in L2 than in L3. Small particle MRT in L2 was longer than MRT for liquid, though correlations were high (r = .74 to .86, P less than .01). Large particles were retained longer than small particles. Particle size, morphology, and percentage of stem in the forage influenced the mean retention time. Use of representative samples of all fractions fed to measure retention times may lead to a better understanding of ruminal function and to improvements in explanations of intake regulation, because marked fractions do not behave identically for all forages.