Three diets were fed to 6 horses in a 3 × 6 Latin rectangle experiment to determine if dietary fibre could improve hydration status in response to dehydration with frusemide. Frusemide was used to simulate dehydration from exercise-induced sweat loss. Diets contained similar dry matter (DM), energy, protein and electrolyte content, but differed in total dietary fibre (TDF) and/or soluble fibre (SDF). The 3 diets were: 1) HIGH-HIGH (high TDF, high SDF); 2) HIGH-LOW (high TDF, low SDF); and 3) LOW-LOW (low TDF, low SDF). In each 10 day period, water intake and faecal moisture content were assessed on Day 7. On Day 10, feed and water were withheld and horses were dehydrated with frusemide (1 mg/kg bwt, i.m.). Plasma volume (PV), plasma total protein (TP), packed cell volume (PCV) and plasma electrolyte concentrations were determined before and after frusemide administration. Water consumption was greater (P<.05) when horses received diets high in TDF. Faecal moisture content was greatest (P<.01) when horses received the HIGH-HIGH diet. The decline in PV and the rise in plasma TP concentration following frusemide administration were similar for all diets. When horses received the HIGH-HIGH diet, they had a greater (P<.05) bodyweight before dehydration and lost more (P<.05) bodyweight in response to frusemide. A greater loss of bodyweight in response to frusemide without a proportional loss of PV when horses received the HIGH-HIGH diet suggests that a diet high in SDF may provide the horse with a source of dispensable water in the hind gut during dehydration.