The aim of this research is to determine the bacteriological quality of the irrigation canal from Loskopdam, the two rivers that feed it and vegetables (broccoli and cauliflower) in Mpumalanga, South Africa and also to predict the presence of selected bacterial pathogens in irrigation water and on vegetables with logistic regression analysis. Water and vegetable samples were examined for the presence of total coliforms, faecal coliforms, Escherichia coli, Listeria monocytogenes, Salmonella sp., Enterococcus, Staphylococcus aureus, aerobic sporeformers, anaerobic sporeformers and aerobic colony counts were done. Apart from bacterial analysis, the following physico-chemical tests: temperature, pH, turbidity and COD were determined in water samples. The average COD and turbidity in the two rivers and the canal were higher than WHO and SA water guidelines. Sampling and analyses were done for a period of 12 months. Levels of faecal coliforms and E. coli were higher than the WHO standard. S. aureus, Intestinal Enterococci, Salmonella, L. monocytogenes were recovered from the two rivers and the canal. Apart from L. monocytogenes which was not recovered from cauliflower, all bacterial pathogens recovered from the surface water were recovered from the vegetables. 2Practical application: These results show that the rivers in Mpumalanga may contribute to the contamination in the irrigation canal which may be a possible preharvest source of contamination of broccoli and cauliflower, which may in turn constitute a health risk to consumers. Logistic regression analysis of the sampled data showed that COD was statistically reliable to predict L. monocytogenes, turbidity reliable to predict Intestinal Enterococci and faecal coliform and coliform reliable to predict Salmonella in irrigation water. Aerobic colony count (ACC) was statistically significant for the prediction of the three pathogens in vegetables.