In many farm animal species, the relationship to humans affects their welfare considerably. But until now, on-farm studies on fattening bulls have been limited. A feasible, reliable methodology for assessing responses of bulls to humans Would be helpful for large scale surveys on this topic. Measuring avoidance distance at the feeding place (ADF) to assess animals' relationship to humans was shown to be a feasible and stable measure in dairy cow herds. So, the main objective of this study was to investigate reliability aspects of measuring avoidance distance at the feeding place towards an unknown experimenter on bull fattening farms. Moreover, we were interested in the potential confounding factor age of animals, and, on a preliminary level, a possible relationship of responsiveness of bulls to farmers' attitudes. With regard to reliability, between-experimenter repeatability - the repeatability of ADF when carried out by different experimenters within short time - was evaluated by two experimenters in a balanced order with 602 fattening bulls on 10 farms. The experimenter was both the stimulus person and the person collecting the data. Between-experimenter repeatability was analysed at an individual and at farm level. We recorded the median and mean values of avoidance distances [ADF median, ADF mean] per farm and the percentages of bulls accepting to be touched [ADF % touch] or showing avoidance distances greater than 20 cm [ADF % > 0.2 m]. Inter-observer reliability, based on individual observations recorded by an experimenter and an observer simultaneously, was tested on six farms. Farmers were asked to complete a questionnaire in order to assess their attitudes (behavioural beliefs and affective attitudes) towards interacting with bulls. Generally, inter-observer reliability for ADF was high (r(s) > 0.9, n = 288/297). At an individual level, moderate between-experimenter repeatability could be attained (r(s) = 0.6, n = 469). At farm level (n = 10), Spearman rank correlation coefficients (r(s)) for between-experimenter repeatability ranged from >0.4 to 0.7. A low negative correlation was found between ADF and age at an individual level (r(s) = -0.14, P = 0.015, n = 320). Despite the rather low range of avoidance distances observed between farms, ADF correlated significantly with some of the farmers' attitudes. ADF was lower in case of farmers expressing more positive affective attitudes towards contacts with the animals (significantly so for the measures ADF mean and ADF % > 0.2 m [P < 0.05]). In conclusion, the use of the avoidance distance test at the feeding place (ADF) is promising for assessing the responses of fattening bulls to humans due to its feasibility, reliability and sensitivity both at an individual and at farm level. Based on a limited number of farms, our first exploration of the relationships between farmers' attitudes and responses of bulls to humans in the home environment suggests possible links that require further investigations. (C) 2009 Elsevier B.V. All rights reserved.