The impact of the human-animal relationship on hen welfare was examined at 29 laying sheds in Australia and the USA. Fear of humans was assessed by counting the number of birds that remained at the front of the home cage, or had their head out of the home cage, during a human approach test. A principal components analysis identified two components from these data, labelled the 'Forward Score' and the 'Heads Out Score'. A high value for either score indicates low fear of humans. Stockperson behaviour in each shed was studied over 2 days by recording the following behaviours (frequency of behaviours/m of aisle length, with frequency of behaviour recorded using a 5 s bout interval): Visual presence in the aisle (Visual); approaching the cages (Approach); physical contact with the cage (Contact); hands inside the cage (Entry), touching the birds (Handle), and human-made noise (Noise). A composite variable (Near Cage) was created by summing 'Approach', 'Contact', 'Entry' and 'Handle'. Stockperson speed of movement (SOM, m/s) and the time spent stationary (Time Stationary, s) were also recorded. A partial correlation analysis was conducted with country partialled out due to the large differences in bird and stockperson behaviour between Australia and the USA. The 'Heads Out Score' was positively correlated with the stockperson behaviours 'Near Cage' (r = 0.48, P < 0.05); 'Contact' (r = 0.40, P < 0.05), and 'Time Stationary' (r = 0.38, P < 0.05), whilst 'Noise' was correlated negatively with both the 'Forward Score' (r = -0.65, P < 0.01) and 'Heads Out Score' (r = -0.59, P < 0.01). These results indicate that bird welfare could be improved by increasing the number of human-animal interactions that occur in close proximity to the cages, and by reducing the amount of noise that stockpeople make in the laying shed.