Studies on the welfare of carriage horses have been addressed during working activity; however, their nocturnal resting behavior remains poorly explored. The aim of our study was to quantify the frequency of negative, positive, and neutral facial expressions (FE) of carriage horses used in tourism displayed under nocturnal resting conditions. Observations were carried out in the tropical city of Mérida, México, from January to March 2021. Nine crossbreed horses, including 3 mares and 6 geldings (400 ± 50 kg, 9-20 years old), were used. Prior to FE observations, an evaluation of resting locations was performed with regard to available surface area, movement restriction, floor and roof conditions, and visual and auditory contact between horses. A 40-point veterinary health examination was also performed on all horses. FE were identified based on the Horse Grimace Scale and EquiFacs coding system. Occurrences of stiffly backwards ears EAD103, orbital tightening AU145, tension above the eye area AU5, prominent strained chewing muscles AU18, strained mouth and pronounced chin AU24+17, and strained nostril and flattening of the profile EAD103+AUH13 were considered negative; ears forward EAD101 and nostril dilation AD38 were considered positive; and right ear forward EAD101R and without movement (WM) were considered neutral. All FE were photographed with a hidden camera (GoProHero3+, 12 megapixels) every minute for at least 8 hours overnight immediately following a randomly selected working day. Good resting location scores were fulfilled by 100% of horses except for good floor and roof conditions (66.6%). Physical health evaluations resulted in 90% of horses with good health scores. The most frequent FE per hour were WM (27.0 ± 5.6), EAD101 (12.8 ± 5.6), and EAD103 (8.9 ± 5.8). Statistical differences (P < 0.0001) were found for mean comparison test between neutral (15.7 ± 3.1 FE/h), positive (6.6 ± 2.9 FE/h), and negative (2.4 ± 1.3 FE/h) FE/h. No significant differences were found for the frequency of neutral, positive, or negative FE/h according to age, body condition score, or health status of the animals, nor according to floor and roof condition of resting locations. The higher frequency of neutral and positive FE compared to negative FE is consistent with the good health scores and generally good conditions of resting locations in the large majority of the horses observed.