Urine samples with trimethoprim-resistant or trimethoprim-sensitive Gram-negative bacteria and samples with no bacterial growth (NG) were identified. Age-sex matched community controls were generated with each trimethoprim-resistant case. These four groups were evaluated for exposure. Prior trimethoprim use was significantly more common in the trimethoprim-resistant group when compared with the trimethoprim-sensitive or the NG group. Prior hospitalization was significantly less common in the trimethoprim-resistant than the trimethoprim-sensitive group, but not with the NG group. Prior oestrogen exposure was associated with trimethoprim resistance. There were no associations found for diabetes or prior corticosteroid exposure. Community controls were found to be inappropriate controls for the study of trimethoprim-resistant bacteria in urine samples.