We examine whether the relaxation of constraints on patient choice of hospital in the English National Health Service in 2006 led to greater changes in mortality for hospitals which faced more rivals before the choice reform. We use patient level data from 2002 to 2010 for three high volume emergency conditions with high mortality risk: acute myocardial infarction (AMI) (288,279 patients), hip fracture (91,005 patients), stroke (214,103 patients). Since mortality risk varies by sub-diagnoses of AMI and stroke we include indicators for sub-diagnoses in the covariates. We also allow for the effect of covariates on mortality to differ before and after the 2006 choice reform. We find that the choice reform reduced mortality risk for hip fracture patients by 0.62% (95% CI: 1.22%, 0.01%), compared with the 2002/3-2010/11 mean of 3.5%, but had statistically insignificant negative effects for AMI and stroke. The reform also had heterogeneous effects across AMI and stroke sub-diagnoses, reducing mortality for 3% of AMI patients and 21% of stroke patients. The reduction in hip fracture mortality was greater for more deprived patients. Policies to increase competition and give patients greater choice are likely to have heterogeneous effects depending on details of patient case mix and market conditions.