We aimed to verify whether a low number of relevant animal-based indicators was able to discriminate 33 semi-intensive (grazing during the day and confinement during the night with access to an outdoor paddock; S-INT) and 8 intensive farms (permanent confinement with access to an outdoor paddock; INT) located in the Mexican semi-desert. In addition, we implemented the resource-based assessment scheme Animal Needs Index (ANI) with the identified animal-based indicators to compare the overall level of welfare in INT and S-INT. In particular, we used a protocol made up of 2 parts. The first comprised 4 evaluation sheets (locomotion, flooring, environment, management) and resource-based indicators derived from ANI, and the second one comprised a set of validated animal-based measures focusing on physical conditions and clinical signs of disease derived from the Animal Welfare Indicators scheme and reported in 2 additional sheets. The scoring system was also derived from ANI, with partial scores for each sheet to be summed to obtain the total score. A total of 1,116 dairy goats were assessed. All the observations and recordings were performed by an expert veterinarian evaluator assisted by an auxiliary, and longevity was retrieved from the farm records. The prevalence of animals displaying dirtiness, ocular discharge, abscesses, and claw overgrowth were higher in INT than in S-INT. Disbudding was routinely performed in INT only. Therefore, scurs, indicating improper disbudding, were recorded only in INT. In addition, the longevity of goats raised in S-INT was higher than in INT. Conversely, the prevalence of goats affected by anemia (i.e., FAMACHA scores >2) or lean (i.e., body condition score <2) tended to be higher in S-INT than in INT. No significant differences between the 2 groups of farms were detected for wounds, nasal discharge, integument alterations, fecal soiling, uterine prolapse, and subclinical mastitis. The results obtained using only animal-based measures were confirmed when resource-based variables were also included in the assessment, as 3 out of 6 sheets of the evaluation scheme (i.e., flooring, environment, and health-physical conditions) were scored higher in the S-INT than in the INT. As a consequence, the total score was also higher for S-INT than for INT. We conclude that the selected set of validated animal-based measures was able to discriminate between farms from different production systems. In particular, higher welfare levels were observed in S-INT farms, where the animals were allowed to spend most of the day on natural pasture, compared with INT farms, where the animals were constantly confined. Nevertheless, a certain degree of improvement should also be promoted in terms of anemia and body condition in S-INT farms.