In a cross-sectional study with elements of cohort design, we investigated relationships between animal handling and stress-related animal behaviours in connection with slaughter at two Swedish slaughter plants: a newly started small-scale on-farm mobile abattoir and a relatively large-scale stationary slaughterhouse. To the stationary plant, the animals were transported on average 99 km from farms, and one third of these animals spent one night in lairage before slaughter. Data were collected during processing of 298 animals at both plants during one year. Stockperson actions and animal behaviours were observed in the driveways (2.4 to 7.3 m long) to the stun box. Data on season, hour of day, air temperature, animal breed, animal category, animal age, carcass weight and stockperson category (plant or farm) were also collected. We used Spearman rank correlation, principal-component analysis and confirmatory factor analysis to develop a final generalised structural equation model, which contained six variables that represented observed animal backing; turning; slipping; eliminating; vocalising; and violent behaviour, regressed on a latent variable representing animal stress level. Stress level and animal behaviours were also regressed on five variables representing observed stockperson actions (touching, patting or hitting with hand; touching, patting or hitting with a tool; pushing; tail-twisting; and prodding with an electric goad) and on background variables including plant identity. The animal behaviours were modelled as counts per driveway length, clustered on farm identity. Marginal effects of stockperson actions and predictive margins of plants were calculated. The animals displayed backing and violent behaviour (kicking, goring or violent fighting) at significantly lower frequencies at the mobile plant than at the stationary one. In general, stockperson actions were significantly positively associated with animal behaviours, i.e. increased actions were associated with more frequent behaviours. Stockperson moving with tool, pushing and using electric goad were indirectly significantly associated with all animal behaviours via animal stress level. This study shows the importance of adequate cattle handling to limit pre-slaughter stress.