The study objectives were to identify potential associations between reported antimicrobial use (AMU) practices and antimicrobial resistance (AMR) of fecal and environmental Salmonella spp. isolates (n = 322 isolates) recovered from 60 Alberta finishing swine farms, and to estimate the amount of pen and farm level variation in AMR. The AMU data were collected through a questionnaire. Separate multilevel logistic regression models were built for six antimicrobials with prevalence of resistance >or=5% using the Generalized Linear Latent and Mixed Model (GLLAMM) procedure. In-feed use of tylosin in finishers was associated with increased odds of resistance in Salmonella isolates to ampicillin (OR = 61.56), streptomycin (OR = 11.70), and multiple antimicrobials (OR = 4.90). Injectable penicillin use in growers was associated with decreased odds of resistance in Salmonella isolates to streptomycin (OR = 0.06), kanamycin (OR = 0.03), and multiple antimicrobials (OR = 0.12). Injectable penicillin use in finishers was associated with decreased odds of resistance in Salmonella isolates to ampicillin (OR = 0.007) and chloramphenicol (OR = 0.04). Overall, these results indicate that AMU in pig production is inconsistently associated with AMR in Salmonella from finishing swine. Variation in AMR prevalence of Salmonella isolates of swine was moderate to high at pen and farm levels for most antimicrobials suggesting that interventions at the pen and farm levels might be beneficial in reducing the emergence of AMR Salmonella in swine populations.