Fifty two children with insulin dependent diabetes mellitus were randomised to receive human isophane or lente insulin preparations in combination with soluble insulin in a double blind trial. Patients were seen every two months, and crossed over after four months of treatment. Control assessed by glycated haemoglobin was significantly lower in children on human isophane insulin, but fasting blood glucose and fructosamine concentrations and the number of episodes of hypoglycaemia were similar on both regimens. In five children on twice daily insulin regimens, insulin profiles throughout a 24 hour period demonstrated greater variability on lente compared with isophane insulin despite identically administered insulin doses. A questionnaire completed at the end of the study showed that two thirds of the children and/or their parents preferred the isophane insulin, and they gave perceived improvement of metabolic control as the major reason for their choice.