Patient safety is a priority in healthcare today. Good clinician numeracy in doctors contributes to patient safety, since it is essential for accuracy in prescribing, and in data interpretation. Evidence, however, suggests that although doctors are assumed to be highly numerate, they often make errors in drug dose calculation and struggle to interpret medical statistics. Having developed a new assessment measure, the Medical Interpretation and Numeracy Test (MINT), we describe its use to evaluate clinician numeracy in 135 recently qualified doctors in the UK ("foundation trainees," i.e., in their first two years post-graduation). The maximum possible test score was 43; the range of scores was 14 -42 (33 -98% correct), with an interquartile range of 29 -38 (67 -88%). Mean score was 32.76 (76%), with a 95% confidence interval of 31.6 to 33.9. Drug dose calculation errors were common, and potentially hazardous. Two thirds of participants had difficulty with simple data interpretation tasks designed for patients. Almost a quarter could not select the better of two treatment options when data were presented as relative risk reduction, absolute risk reduction or number needed to treat. Our study suggests that a significant proportion of medical graduates have poor clinician numeracy. Such doctors may harm patients by making mistakes in drug dose calculation and through flawed medical decision making. While further research is needed to investigate clinician numeracy in doctors, we believe that there is sufficient evidence to incorporate clinician numeracy into undergraduate and postgraduate medical education programs as a matter of patient safety.