The selection of the best control group is a major concern in the design of any case-control study, because of its enormous implications on both interval validity and cost. We compare the dietary habits of hospital and neighbourhood control groups which had both been used in a case-control study on bladder cancer. The analysis is based on 485 matched sets. No differences were observed between the two groups in terms of nutrient consumption nor in consumption of specific foods or food groups. Alcohol consumption represents an exception. This may be attributed to the use of incorrect exclusion criteria given that alcohol consumption was not one of the major determinants of the original study. We conclude that, provided correct exclusion criteria are used, hospital controls may be a suitable as neighbourhood controls in the assessment of dietary risk factors in case-control studies in Spain.