Using Bayesian network analysis, this cross-sectional study aimed to identify the conditional probability among dairy farm practices, cow characteristics, bacteriological culture results, and antimicrobial susceptibility test results of milk from dairy cows with clinical mastitis in western Thailand. Data associated with risk factors and clinical signs were collected using a structured questionnaire that was administered to 34 small dairy holders. In total, 100 quarters of milk samples from 100 cows were used for Bayesian network analysis. Conditional probability results showed that the following variables had the highest probabilities relevant to the occurrence of clinical mastitis pathogens: parity, concrete and rubber floor, hand stripping after using machine milking, dry cow therapy, and routine cleaning of milking machines. These variables were associated with the first four highest posterior probabilities of the occurrence of Streptococcus spp. (16.68%; reachable range or the minimum and maximum posterior probability values for the occurrence of Streptococcus spp., 15.45%–17.91%), Staphylococcus spp. (11.87%; reachable range, 11.06%–12.67%), Escherichia coli (7.53%; reachable range, 6.95%–8.17%), and Streptococcus dysgalactiae (7.28%; reachable range, 6.73%–7.83%), which were the most frequently isolated pathogens. Conditional probability results indicated these pathogens were most sensitive to amoxicillin/clavulanic acid (80.58%) and cloxacillin (64.28%). Most pathogens were resistant to penicillin G (40.37%). In this study, Bayesian network analysis revealed several clinically significant risk factors of mastitis associated with various pathogens and farming characteristics. Simple statistics could not provide sufficient information for the successful control of mastitis. In contrast, through in-depth data analysis, Bayesian networks could identify risk factors in various situations, hence providing information that will be crucial to help farmers reduce the cost of farming.