Animal welfare has become a matter of an emotional public debate, leading to the necessity of developing an objective measurement tool. This is also valid for the horse industry which has been criticized intensively in Germany during the last few years. In 2015, the Animal Welfare Indicators (AWIN) protocols were developed amongst others for horses mainly based on animal-based indicators. This is of great interest as German law demands farmers to carry out a regular self-assessment regarding their husbandry by animal-based indicators. However, a detailed verification on farm of the included animal-based indicators has up to date not been accomplished. Therefore, the present study aims at testing feasibility as well as interobserver and retest reliability of the protocol in its on farm use on horse farms in Germany. Furthermore, it is analyzed whether the division into two levels, i.e. a timesaving first level assessment as overview and only if potential welfare issues are detected a more time-consuming second level, is suitable. In general, the protocol is feasible as no expertise knowledge or special equipment is required. In order to achieve comparable results, a training of observers is recommended. The time needed varies between 5-15 min per horse. The number of observed horses per farm is dependent on the total farm size. In our study, the total time to complete a protocol assessment was 2-8 h. The division into two levels in the protocol for horses was only suitable for some of the indicators (ocular discharge, coughing, wounds on the back and at the withers as well as swellings). As data collection continues, reliability assessment and therewith a final evaluation of each included animal based indicator will be completed. Based on this, a reasonable checklist for the regular self-assessment will be developed.