OBJECTIVE: To assess whether patients from non-white ethnic groups wait longer than white patients for elective in-patient admissions at St Mary's Hospital in London. METHODS: Patients who came off the waiting list for an elective inpatient admission between 1 April 2000 and 31 March 2001 were selected. A multivariable log linear model was developed to assess geometric mean waiting times for Black, Asian, Other and Missing ethnic groups compared to the White group, adjusted for age, sex, urgency and distance. RESULTS: Caution is needed in interpreting results, as a large number of patients had no usable ethnic code. There was no strong evidence that waiting times for ethnic groups were systematically different than for the White group. However, there was some evidence that white patients waited longer for a coronary arteriography than patients in other ethnic groups. This was partially explained by age, sex, clinical urgency and residential distance from St Mary's. CONCLUSIONS: The large proportion of patients with no usable ethnic code, lack of robust methods for case-mix adjustment and multiple ethnic categories makes analysis methodologically difficult. Regular and informative analysis of ethnic coded data is a necessary step in improving the accuracy and completeness of coding.