There is community concern about the treatment of farm animals post-farm gate, particularly animal transport and slaughter. Relationships between lamb behavioural and physiological variables on farm, stockperson, dog and lamb behavioural variables pre-slaughter and plasma cortisol, glucose and lactate in lambs post-slaughter were studied in 400 lambs. The lambs were observed in three behavioural tests, novel arena, flight distance to a human and temperament tests, before transport for slaughter. Closed-circuit television video footage was used to record stockperson, dog and lamb behaviour immediately before slaughter. Blood samples for cortisol, glucose and lactate analyses were collected on farm following the three behavioural tests and immediately post-slaughter. The regression models that best predicted plasma cortisol, glucose and lactate concentrations post-slaughter included a mixture of stockperson and dog behavioural variables as well as lamb variables both on-farm and pre-slaughter. These regression models accounted for 33%, 34% and 44% of the variance in plasma cortisol, glucose and lactate concentrations post-slaughter, respectively. Some of the stockperson and dog behaviours pre-slaughter that were predictive of the stress and metabolic variables post-slaughter included the duration of negative stockperson behaviours such as fast locomotion and lifting/pulling lambs, and the duration of dog behaviours such as lunging and barking at the lamb, while some of the predictive lamb behaviour variables included the durations of jumping and fleeing. Some of the physiological and behavioural responses to the behavioural tests on farm were also predictive of the stress and metabolic variables post-slaughter. These relationships support the well-demonstrated effect of handling on fear and stress responses in livestock, and although not direct evidence of causal relationships, highlight the potential benefits of training stockpeople to reduce fear and stress in sheep at abattoirs.