The variability of animal-based parameters was studied within a population of 41 farrow-to-finish farms. Data were collected during three visits, each corresponding to a different season within a two-year period. The largest between-farm variability was observed for stereotypic behaviour by pregnant sows, and for skin, ear and tail lesions, dirtiness and respiratory problems in growing pigs. Relationships with housing and management parameters were established to formulate advice on how to improve pigs' welfare. Group-housed sows performed less oral stereotypic behaviour than individual housed sows (18.7 versus 44.1%), but a higher proportion of skin lesions was observed in group-housed sows (15.4 versus 2.0%). Prevalence of tail-biting behaviour varied between 0 and 21%. The risk for tail biting was higher in cases of reduced levels of floor space per pig, and ear-biting behaviour occurred more often when tails were docked short. Coughing was not correlated directly with the occurrence of lung lesions, but the risk was higher in instances of reduced space availability per pig. Farms could be ranked according to these welfare parameters, ie either according to the score of each individual parameter or based on the summation of all scores. Hence, welfare status was defined in relation to farm-specific information, allowing formulation of advice on housing and management to ultimately improve pig welfare through the matching of a predefined benchmark.