Morphological characteristics of the forestomach, as well as reports of a natural diet that mostly excludes monocots, suggest that dikdiks (Madoqua spp.), among smallest extant ruminants, should have a ‘moosetype’ forestomach physiology characterised by a low degree of selective particle retention. We tested this assumption in a series of feeding experiments with 12 adult Phillip's dikdiks (Madoqua saltiana phillipsi) on three different intake levels per animal, using cobalt-EDTA as a solute marker and a ‘conventional’ chromiummordanted fibre (b2 mm; mean particle size 0.63 mm) marker for the particle phase. Body mass had no influence on retention measurements, whereas food intake level clearly had. Drinking water intake was not related to the retention of the solute marker. In contrast to our expectations, the particle marker was retained distinctively longer than the solute marker. Comparisons with results in larger ruminants and with faecal particle sizes measured in dikdiks suggested that in these small animals, the chosen particle marker was above the critical size threshold, above which particle delay in the forestomach is not only due to selective particle retention (as compared to fluids), but additionally due to the ruminal particle sorting mechanism that retains particles above this threshold longer than particles below this threshold. A second study with a similar marker of a lower mean particle size (0.17 mm, which is below the faecal particle size reported for dikdiks) resulted in particle and fluid retention patterns similar to those documented in other ‘moose-type’ ruminants. Nevertheless, even this smaller particle marker yielded retention times that were longer than those predicted by allometric equations based on quarter-power scaling, providing further support for observations that small ruminants generally achieve longer retention times and higher digestive efficiencies than expected based on their body size.