To evaluate the use of the caudal vena cava collapsibility index (CVCCI) as a predictor of fluid responsiveness in hospitalized, critically ill dogs with hemodynamic or tissue perfusion abnormalities. Retrospective observational study. Private referral center. Twenty-seven critically ill, spontaneously breathing dogs with compromised hemodynamics or tissue hypoperfusion. None. The electronic medical records were searched for dogs admitted for any cause, from August 2016 to December 2017. We included dogs with ultrasound measurements of: CVCCI, performed at baseline; and velocity time integral (VTI) of the subaortic blood flow, carried out before and after a fluid load. CVCCI was estimated as: (maximum diameter-minimum diameter/maximum diameter) × 100. Dogs in which VTI increased ≥15% were considered fluid responders. The CVCCI accurately predicted fluid responsiveness with an area under the receiver operating characteristic curve of 0.96 (95% CI, 0.88 to 1.00). The optimal cut-off of CVCCI that better discriminated between fluid responders and nonresponders was 27%, with 100.0% sensitivity and 83.3% specificity. At baseline, fluid responders had lower VTI (5.48 [4.26 to 7.40] vs 10.61 [7.38 to 13.23] cm, P=0.004) than nonresponders. The basal maximum diameter of the caudal vena cava adjusted to body weight was not different between responders and nonresponders (0.050 [0.030 to 0.100] vs 0.079 [0.067 to 0.140] cm/kg, P=0.339). The increase in VTI was related to basal CVCCI (R=0.60, P=0.001). Bland-Altman analysis showed narrow 95% limits of agreement between measurements of CVCCI and VTI performed by different observers or by the same observer. The results of this small cohort study suggest that CVCCI can accurately predict fluid responsiveness in critically ill dogs with perfusion abnormalities. Further research is necessary to extrapolate these results to larger populations of hospitalized dogs.