Summary objective To identify risk factors for urinary tract stones in Trinidad and Tobago. methods A consecutive series of patients presenting to institutions in Trinidad for the management of proven urinary tract calculi was interviewed by questionnaires designed to obtain data on age, gender, ethnicity, occupation, stone location, a family history of stone disease, a past history of certain medical diseases and a semiquantitative estimate of the magnesium intake in food, ‘over the counter’ drugs and drinking water. An equivalent number of patients attending the same institutions for follow-up and verified from hospital records as having a previous radiological diagnosis of urinary tract stones, as well as a group of asymptomatic members of the community working in the same area, underwent the same interview. Chi-square,anovaand Kruskal–Wallis tests were used to examine differences between the groups. Multiple logistic regression analysis was used to determine persistence of the significance of these differences after controlling for confounding variables. results Data sufficient for analysis were obtained for 122 previous and 102 prospective patients and 102 controls. The mean age of the patients was 32 years. The ratio of males to females was 0.9–1.8:1 for the affected groups, but significantly more males than females had calculi in the lower urinary tract (19%vs. 6%; P = 0.004). More patients (30%) than controls (7%) gave a positive family history of urinary tract stone disease (P < 0.001). Affected persons had a lower dietary magnesium intake (P = 0.003), which accounted for a significantly lower total magnesium intake (P = 0.02). Logistic regression analysis of the variables studied indicated that independent predictors of the disease were a positive family history (P = 0.001), total magnesium intake (P = 0.001) and age (P < 0.001). conclusion A low magnesium intake and a positive family history are highly predictive of urinary tract calculi in this population.