We found good evidence that intensive insulin therapy improves glucose control and reduces microvascular complications compared with conventional treatment in people with uncomplicated and poorly controlled type 1 diabetes mellitus, although absolute risk reductions associated with treatment are likely to be greater in those with higher baseline risk. However, intensive treatment is associated with increased risk of severe hypoglycaemic episodes and ketoacidosis. We found limited evidence to compare insulin pumps with multiple insulin injection. Limited evidence suggests that insulin pumps may improve glucose control but increase the risk of ketoacidosis compared with multiple injection. We found no evidence to compare insulin pumps and multiple injections for chronic diabetic complications.