We assessed the long term efficacy of Novopen as a form of insulin administration. Records were obtained on 48 patients who were treated with Novopen between January '86 and October '88. Six patients were excluded due to insufficient data. The study group of 42 patients comprised 22 females and 20 males of average age 33 years (range 17-66). Mean Hb.A1 rose from 10.6% to 12.1% after Novopen therapy, a rise of 14.1%. This rise is both clinically and statistically significant (p less than 0.001; 99% confidence limits 0.59-2.78). Increases in weight and insulin dose were also noted, but did not reach statistical significance. The majority of patients felt Novopen was superior to twice daily insulin in terms of ease of administration (81%) and flexibility of lifestyle (95%), and all who were using Novopen wished to continue with it. More than 50% of patients admitted to altering their dietary habits while using Novopen. Despite continuing patient satisfaction with this form of insulin administration, its long-term use may be associated with sub-optimal metabolic control.