Acute stressors elicit behavioural and physiological adaptations that promote survival, including increased vigilance, reduced exploratory behaviour, and decreased appetite. Complex neural circuits contribute to these processes, with the caudal nucleus of the solitary tract (cNTS) and its resident glucagon-like peptide-1 (GLP1) neurons playing a key role1. GLP1 neurons are activated following stress2,3 and mediate stress-induced suppression in feeding4, but the inputs which drive their activation in response to stress are unknown. To identify stress-activated direct inputs to GLP1 neurons we performed monosynaptic retrograde tracing. Glu-cre/tdRFP mice (n=9, 4 females, 5 males) were anaesthetised using isofluorane (1.5-3%) and stereotaxically injected with a 1:1 cocktail of AAV8-CAG-FLEX-RabiesG and AAV5-EF1a-TVA:mCherry (200nl) targeted to the cNTS followed three weeks later by injection of EnvA-RABV-deltaG-GFP (400nl). Seven days later mice were exposed to 30mins restraint stress (or left undisturbed) and transcardially perfused (4% formaldehyde) after 90mins. Coronal sections (40µm) were immunolabelled for GFP and cFOS to identify recently activated, RABV-infected neurons. The paraventricular nucleus of the hypothalamus (PVN) provided dense monosynaptic input to GLP1 neurons, and acute stress increased the number of activated RABV-GFP-positive neurons by 27.1 percentage points [95% confidence interval (95%CI): 16.1, 47.8 percentage points; p=0.036, Student’s T-test). To establish whether PVN input is necessary for stress-induced activation of GLP1 neurons, we used an intersectional approach to selectively inhibit cNTS-projecting PVN neurons. Wildtype C57BL/6 mice (n=13 males) were injected with rgAAV-hSyn-Cre into the cNTS followed by AAV8-hSyn-DIO-hM4Di:mCherry (n=5) or AAV1-CAG-FLEX-EGFP (n=8) into the PVN. Injection of the ligand, CNO (2mg/kg), significantly increased 2h dark-onset food intake in hM4Di-expressing mice by 0.27g (95%CI: 0.19, 0.326; p=0.029), but not in control mice [0.154g (95%CI: 0.324, 0.00625); p=0.11; Sidak multiple comparisons]. Furthermore, while acute stress significantly suppressed food intake in control mice by 0.265g (n=5 males; 95%CI: 0.16, 0.36; p=0.005), mice in which cNTS-projecting PVN neurons were inhibited (n=4) displayed no reduction in chow intake [effect size: 0.055 (95%CI: -0.01, 0.135g); p=0.57; Sidak multiple comparisons]. As a terminal procedure, mice were injected with CNO (2mg/kg), exposed to acute stress, and transcardially perfused 90mins later. Coronal brain sections (35µm) were immunolabelled for cFOS, GLP1, and GFP or mCherry. Chemogenetic inhibition of cNTS-projecting PVN neurons significantly attenuated the number of cFOS-immunoreactive neurons in the cNTS (p=0.0041, Student’s T-test) as well as the percentage of cFOS-immunoreactive GLP1 neurons following acute stress (p=0.033, Student’s T-test). Finally, we determined the molecular phenotype of these cNTS-projecting PVN neurons using RNAscope in situ hybridization and immunolabelling. Approximately half of identified cNTS-projecting PVN neurons expressed Crh mRNA (49.0±1.4%), while only 3.9±1.1% expressed oxytocin (n=3). We also confirmed Crh expression by a subset of RABV-labeled PVN neurons that provide synaptic input to GLP1 neurons. These findings reveal a hypothalamic-hindbrain pathway that mediates stress-induced activation of cNTS neurons and feeding suppression. The experimental protocols were approved by the FSU Institutional Animal Care and Use Committee, and were consistent with the US Public Health Service’s Policy on the Humane Care and Use of Laboratory Animals and the NIH Guide for the Care and Use of Laboratory Animals.